JP2011188368A - Encoding apparatus, decoding apparatus and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an encoding apparatus for compressing an image by converting a color space formed with a plurality of component signals into a non-correlated color space, a decoding apparatus therefor, and a program. <P>SOLUTION: An encoding apparatus (10) includes: a non-correlating converter (101) which receives input of a signal element of each of component signals in a predetermined color space and performs non-correlating conversion; and an encoder (112) for applying encoding processing to the non-correlated signal to generate an encoded signal. Information of a non-correlated matrix used for non-correlating conversion is generated as a non-correlating parameter. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to compression coding of an image, and in particular, uses a correlation between color signals composed of a plurality of component signals, and converts the color space formed by the plurality of component signals into an uncorrelated color space to compress the image. The present invention relates to an encoding device, a decoding device thereof, and a program thereof.

  In the currently used image compression encoding system (MPEG-2, AVC / H.264, JPEG, etc.), the correlation of the time or spatial direction of the image is used, and this is removed to remove the entire information. It is a method to reduce the amount. In a color image, there may be a correlation between a plurality (usually three colors) of component signals. In MPEG and the like, encoding is performed after converting the RGB imaged by the camera into luminance signals and color difference signals such as YCbCr, thereby removing the correlation between the component signals to some extent.

  As a signal interpolation technique using the correlation between component signals constituting an image signal, a technique for improving the accuracy of a pixel interpolation method in a single-plate primary color image sensor has been reported (for example, see Non-Patent Document 1). Since the image signal interpolation technique in such an image sensor is intended to interpolate RGB signals (R: red signal, G: green signal, B: blue signal) in the RGB color space, signal deterioration due to encoding is assumed. It has not been.

  Further, as a signal interpolation technique focusing on the difference in sampling frequency of YUV signals in the YUV color space, a color difference component interpolation technique for format color images has been reported (for example, see Non-Patent Document 2). In this technique, high-precision interpolation is performed by generating an interpolated signal of a color difference signal (U signal = BY, V signal = RY) using the sampling frequency of the luminance (Y) signal. . The signal interpolation technique that pays attention to such a difference in the sampling frequency of the YUV signal is also intended for the interpolation of the YUV signal, so that signal degradation due to encoding is not assumed.

  These signal interpolation techniques are suitable for interpolation on an image signal before encoding when encoding an image signal by an irreversible encoding method (for example, MPEG-2, H.264, etc.). It is not suitable for interpolation on the image signal after conversion. For example, when a YUV signal is encoded by irreversible encoding processing, the deterioration of the luminance signal propagates to the color difference signal based on the luminance signal as the luminance signal deteriorates. In addition, these signal interpolation techniques are not processes for reducing the degradation of the luminance signal itself, and therefore do not reduce the degradation of the luminance signal.

  In addition, there are various post filters (for example, deblocking filters in H.264, etc.) in order to reduce the degradation of encoding, but these post filters have noticeable degradation of each of the image signal components visually. The processing is performed independently so that the deterioration after encoding of the original image signal cannot be reduced.

Kuno, Sugiura, “Higher accuracy of pixel interpolation in single-plate primary color image sensor”, Journal of the Institute of Image Information and Television Engineers, Vol. 61, no. 7. July 1, 2007, pp. 1006 to 1016 Sugita, Taguchi, "YUV4: 2: 0 format color difference component interpolation method", IEICE Transactions, Vol. J88-A, no. 6, June 1, 2005, pp. 751-760

  However, even a signal converted from a luminance signal such as RGB captured by a camera into a luminance signal and a color difference signal such as YCbCr remains locally, and further compression is possible.

  Further, the deterioration of the orthogonal transform coefficient due to quantization is perceived as deterioration of a pixel such as block distortion or mosquito noise due to inverse quantization and inverse orthogonal transform. In addition, since the degree of deterioration differs for each pixel block, the difference may be a significant disturbance at the boundary between adjacent coding blocks, and block distortion may be detected. In such a case, there is room for further improving the block distortion by using the correlation between the component signals of the moving image.

  An object of the present invention is to provide an encoding device, a decoding device, and a program capable of achieving further compression while improving block distortion.

  Therefore, the inventors do not interpolate or correct each other using the correlation between the component signals, but completely perform the decorrelation between the component signals using the correlation between the component signals, and contribute to the image. We focused on compressing with distinction.

  In the present invention, utilizing the correlation between component signals that has not been sufficiently utilized in the conventional image coding system, the color space formed by two or more types of component signals is converted into an uncorrelated color space, Preferably, the signal amount of the decorrelated component signal whose signal distribution is reduced with respect to the entire signal amount among the decorrelated component signals (hereinafter referred to as decorrelated component signals) is reduced.

  That is, the encoding apparatus of the present invention is an encoding apparatus that compresses an image by converting a color space formed by a plurality of component signals into an uncorrelated color space, and each component signal signal in a predetermined color space A decorrelation conversion unit that inputs an element and executes decorrelation conversion; and an encoding unit that performs an encoding process on the decorrelated signal to generate an encoded signal, the decorrelation The conversion conversion means generates the information of the decorrelation matrix used for the decorrelation conversion as a decorrelation parameter.

  Further, in the encoding device of the present invention, the decorrelation conversion means determines principal components in descending order of eigenvalues using a principal component analysis method, and the signal elements of each component signal are decorrelated with the principal components. The decorrelation parameter is generated by constructing a decorrelation transformation matrix for transforming the signal component of the component signal.

  Further, in the encoding device of the present invention, a predetermined overall signal with respect to one or more decorrelation component signals of the decorrelation component signals immediately after decorrelation by the decorrelation conversion means. It is further characterized by further comprising a signal amount reducing means for reducing the signal amount in accordance with a reduction rate with respect to the amount.

  Further, in the encoding device of the present invention, the signal amount reducing means is for the decorrelated component signal having the smallest signal distribution among the decorrelated component signals immediately after decorrelation by the decorrelation converting means. The amount of signals is reduced by reducing the number of samples.

  Further, in the encoding device of the present invention, the signal amount reducing means is for the decorrelated component signal having the smallest signal distribution among the decorrelated component signals immediately after decorrelation by the decorrelation converting means. It is characterized in that the amount of signal is reduced by applying rough quantization.

  Further, in the encoding device of the present invention, a signal prediction unit that predicts a signal in time or space direction to generate a prediction signal, and a signal axis that is the same between the reference signal used for the signal prediction unit and the prediction signal And a signal axis converting means for converting so as to be.

  In the encoding apparatus of the present invention, a finite number of decorrelation transform matrices are defined in advance as decorrelation transform matrices that can be identified by unique numbers in advance, and the decorrelation transform means includes a principal component analysis method. Auxiliary information that selects the closest one from the finite number of decorrelation transformation matrices that is the closest to the decorrelation transformation matrix obtained in step 3, and associates the unique number of the selected decorrelation transformation matrix with the original combination of component signals It is determined as follows.

  The decoding device of the present invention is a decoding device that decodes an image-compressed signal by converting a color space formed by a plurality of component signals into an uncorrelated color space, and each component signal in a predetermined color space A decorrelation transformation is performed on the signal element of the obtained, and a coded signal subjected to the coding process, and a decorrelation parameter composed of information of the decorrelation matrix used for the decorrelation transformation, Decoding means for decoding the encoded signal; and decorrelation inverse transform means for performing decorrelation inverse transform using the decorrelation parameter and restoring signal elements of the component signal in the original color space; It is characterized by providing.

  Further, in the decoding device of the present invention, the decorrelation inverse transform means performs inverse transform from the signal element of the decorrelation component signal to the signal element of each component signal in the original color space using the decorrelation parameter. A decorrelation inverse transformation matrix is constructed.

  In addition, the decoding apparatus of the present invention has the same signal axis between a signal prediction unit that predicts a signal in time or space direction and generates a prediction signal, and a reference signal used for the signal prediction unit and the prediction signal. And a signal axis converting means for converting as described above.

  In the decoding apparatus of the present invention, a finite number of decorrelation transform matrices are defined in advance as decorrelation transform matrices that can be identified by unique numbers in advance, and the decorrelation inverse transform means is provided on the encoding side. According to the auxiliary information indicating the unique number of the decorrelation transformation matrix determined by selecting the closest one to the decorrelation transformation matrix obtained by the principal component analysis method from the finite number of decorrelation transformation matrices. A decorrelation transformation matrix is specified, and the decorrelation inverse transformation is executed.

  In the present invention, each of the encoding device and the decoding device can be configured as a computer, and each component of the encoding device and the decoding device can be realized by executing a program.

  According to the present invention, by utilizing the correlation between component signals that has not been sufficiently used in the conventional image coding scheme, the signal distribution is preferably compared with the total signal amount of the uncorrelated component signals. In order to reduce the signal amount of the reduced decorrelation component signal, further compression is possible as compared with the conventional image coding method.

It is a block diagram of the encoding apparatus of Example 1 by this invention. It is a block diagram of the decoding apparatus of Example 1 by this invention. It is the schematic which shows the relationship between the signal axis of the component signal of the original color space which concerns on this invention, and the signal axis of a decorrelation component signal. It is the schematic which shows an example of the prediction process between screens / the prediction between screens by the decorrelation component signal which concerns on this invention. It is a block diagram of the encoding apparatus of Example 2 by this invention. It is a block diagram of the decoding apparatus of Example 2 by this invention. It is a block diagram of the encoding apparatus of Example 3 by this invention. It is a block diagram of the decoding apparatus of Example 3 by this invention. It is a control flow in the computer comprised as an encoding apparatus of one Example by this invention. It is a control flow in the computer comprised as a decoding apparatus of one Example by this invention.

  In the encoding device and the decoding device according to the present invention, the image compression is performed by removing the correlation between the component signals of the color image. The original color space to be input may be in any color space, such as RGB, such as camera output, or signals expressed in YCaCb, YUV, Luv, Lab, XYZ, etc. used for transmission or display. In YCaCb, YUV, Luv, etc., the Y or L component signal (luminance signal) is not subjected to decorrelation conversion (matrix coefficient is set to 1, etc.), and other component signals (color difference signals) 2 It can also be configured to remove the correlation between the two component signals. Further, the original color space can be similarly applied to a multi-spectral color space that handles four or more types of component signals (for example, RGBYMC, etc.) in addition to those composed of three types of component signals. it can.

  When removing the correlation between the component signals of the color image, if the original component signal is in the format format of 4: 2: 2 or 4: 2: 0, the number of samples between the component signals per frame is different. The decorrelation conversion is performed by aligning the number of samples in advance. The conversion of the number of samples can average the signal elements of the component signal with the larger number of samples to the same number as the signal elements of the component signal with the smaller number of samples. Since the resolution cannot be reproduced, it is preferable to interpolate the signal elements of the component signal having the smaller number of samples so as to have the same number as the component signal having the larger number of samples.

  An encoding apparatus according to the present invention that encodes an image in a color space composed of a plurality of component signals includes two or more of a plurality of component signals (for example, a C1, C2, and C3 signals) in the original color space. Are generated by performing decorrelation conversion using a decorrelation conversion matrix that removes the correlation between the component signals, for example, P1 signal, P2 signal, and P3 signal. A residual signal (for example, a P1 residual signal, a P2 residual signal, a P3 residual signal) based on a reference signal for spatiotemporal signal prediction (intra-screen prediction or inter-screen prediction) To do. It is preferable to generate a residual signal by making the decorrelation transformation matrix of the decorrelation component signal used when encoding the reference signal the same. That is, in generating the reference signal, after the encoded signal is locally decoded and the component signal before encoding is reproduced, the decorrelation inverse matrix is used to perform the decorrelation inverse transformation by the decorrelation transformation matrix used at the time of encoding. The stored component signal is read out, and a reference signal is generated by performing a decorrelation transform using a decorrelation transform matrix used for a decorrelation component signal for the next encoding. As a result, the residual signal is generated by making the decorrelation transformation matrix of the decorrelation component signal used when coding the reference signal the same, thereby removing the correlation between the component signals and further compressing the image. Can be realized.

  Hereinafter, the encoding apparatus and decoding apparatus of Example 1 by this invention are demonstrated.

  FIG. 1 is a block diagram of an encoding apparatus 10 according to a first embodiment of the present invention. The encoding apparatus 10 according to the first embodiment of the present embodiment includes a decorrelation conversion unit 101 that inputs a plurality of component signals (C1 signal, C2 signal, C3 signal) in the original color space and performs decorrelation conversion. , And compression encoders 103-1, 103-2, and 103-3 that individually encode the residual signals (P1 residual signal, P2 residual signal, and P3 residual signal) of the decorrelation component signal.

  The decorrelation conversion unit 101 receives a plurality of component signals (C1 signal, C2 signal, C3 signal) in the original color space, performs decorrelation conversion, and generates decorrelation component signals (P1 signal, P2 signal, P3 signal) is generated. The generated decorrelation component signals are sent to compression encoders 103-1, 103-2, and 103-3 each having the same component.

  The decorrelation transform uses principal component analysis, for example, to determine principal components in descending order of contribution to the original image (in descending order of eigenvalues) and to determine the signal elements of each component signal in the original color space. It is preferable to generate a decorrelation parameter by constructing a decorrelation transformation matrix that transforms a signal element of a decorrelation component signal composed of the principal component thereof into a signal element. This is to construct a decorrelation inverse transform matrix that inversely transforms the signal elements of the decorrelation component signal into the signal elements of each component signal in the original color space using the decorrelation parameter on the decoding side. Enable. As a result, a new signal system of the decorrelated component signal can be configured with the axis of the contribution ratio, and the magnitude of the signal distribution can be easily discriminated.

  For example, an example in which a decorrelation transformation matrix is obtained by principal component analysis will be described. For simplicity, it is assumed that the signal elements of the component signals (C1 signal, C2 signal, C3 signal) in the original color space are represented by blocks with 10 samples as shown in Table 1, The depth of each pixel value is 0-10.

Such a correlation coefficient r _xy between the component signals shown in Table 1 is given by Expression (1).

  Here, Sxy indicates the covariance between the component signals, and Sx and Sy indicate the deviations between the component signals, and are given by Equations (2) to (4) for the number of samples n, respectively. It is done.

  With respect to Table 1 above, when the correlation coefficient matrix is obtained by Equation (1), the following correlation coefficient matrix is obtained.

  The eigenvalue λ and the eigenvector can be calculated from the correlation coefficient matrix. When the n-th order square matrix is A and the eigenvalues λ are in a finite set (λεC), the eigenvector x is given by Equation (5).

  The necessary and sufficient condition for the existence of the eigenvector x is that the equation (6) is satisfied for the unit matrix E. Therefore, the eigenvalue λ can be obtained by solving the eigen equation of A with the eigenvalue λ as an unknown. And an eigenvector having a magnitude of 1 for each eigenvalue can be obtained.

  For example, the eigenvalues and eigenvectors obtained for Table 1 are expressed as Table 2. In Table 2, principal components (first principal component, second principal component, and third principal component) are determined in descending order of contribution rate (in descending order of eigenvalues), and a decorrelation transformation matrix is constructed. Yes.

  The decorrelation transformation matrix constituting the decorrelation transformation is an array of eigenvectors, and is expressed by equation (7). The decorrelation component signals (P1 signal, P2 signal, P3 signal) corresponding to each principal component (first principal component, second principal component, third principal component) are converted into the original color by this decorrelation transformation matrix. It is obtained by decorrelating a plurality of spatial component signals (C1, C2, and C3 signals).

  In this way, the decorrelation conversion unit 101 receives a plurality of component signals (C1 signal, C2 signal, C3 signal) in the original color space, performs decorrelation conversion, and generates a decorrelation component signal (P1 signal). , P2 signal, P3 signal), and sends the information of the functions constituting the decorrelation transform matrix to the decorrelation transform unit 117, decorrelation parameter storage unit 119, and the like as decorrelation transform parameters.

  The compression encoding units 103-1, 103-2, and 103-3 function in the same manner, and individually encode the decorrelation component signals (P1 signal, P2 signal, and P3 signal). Therefore, the compression encoding unit 103-1 will be typically described.

  The compression encoding unit 103-1 includes a sample number reduction / coarse quantization unit 110, a subtraction unit 111, an encoding unit 112, a local decoding unit 113, an addition unit 114, and a decorrelation inverse conversion unit 115. , An image memory 116, a decorrelation conversion unit 117, an intra-screen prediction / inter-screen prediction unit 118, and a decorrelation parameter storage unit 119.

  The sample number reduction / coarse quantization unit 110 receives, from the decorrelation conversion unit 101, one of the decorrelation component signals obtained by decorrelating the component signals in the original color space. The signal amount is reduced according to the reduction rate with respect to the signal amount, and is sent to the subtraction unit 111 and the intra-screen prediction / inter-screen prediction unit 118. In this example, description will be made assuming that signals are assigned as a P1 signal, a P2 signal, and a P3 signal in order from a signal with a low contribution ratio and a low contribution ratio. FIG. 3 shows the relationship between the signal axis of the component signal in the original color space and the signal axis of the decorrelation component signal according to the present invention. Therefore, the P1 signal is a non-correlated component signal with a low contribution ratio and the lowest signal distribution, and this signal amount is reduced more than the signal amount of the other decorrelated component signals to reduce the subtraction unit. 111 and the intra prediction / inter prediction unit 118. This method of reducing the amount of signal reduces the number of samples of the decorrelated component signal on the axis with a low contribution rate or coarsens the quantization. As a result, the compression rate can be increased while suppressing loss in the overall image quality.

  The subtraction unit 111 is a residual obtained by subtracting the decorrelation component signal obtained from the sample number reduction / coarse quantization unit 110 and the prediction signal of the decorrelation component signal obtained from the intra prediction / inter prediction unit 118. A signal (P1 residual signal) is generated and sent to the encoding unit 112.

  The encoding unit 112 encodes the residual signal (P1 residual signal) obtained from the subtracting unit 111 according to a predetermined encoding procedure, generates a bit stream (P1 residual signal bit stream), and transmits the bit stream to the outside. The encoding unit 112 can be configured as orthogonal transform processing and quantization processing, for example, as in MPEG, and can be configured to perform variable length encoding.

  The local decoding unit 113 decodes the encoded signal obtained from the encoding unit 112 and sends it to the adding unit 114. When the encoding unit 112 is configured as an orthogonal transform process and a quantization process as in MPEG, the local decoding unit 113 performs an inverse transform (an inverse orthogonal transform process and an inverse quantization process).

  The adding unit 114 adds the predicted signal of the decorrelated component signal obtained from the intra prediction / inter prediction unit 118 and the decoded signal obtained from the local decoding unit 113 to obtain a restored signal of the decorrelated component signal. Generated and sent to decorrelation inverse transform section 115.

  The decorrelation inverse transformation unit 115 obtains the decorrelation from the decorrelation parameter storage unit 119 in order to obtain the decorrelation transformation matrix used for the decorrelation component signal at the time of encoding of the encoding unit 112. A parameter is read out, a decorrelation inverse transformation matrix is generated from the decorrelation parameter, a decompressed signal (P1 signal) of the decorrelation component signal obtained from the adder 114, and compression encoders 103-2 and 103 3 is subjected to inverse transformation using a decorrelation inverse transformation matrix on the decorrelated component signals (P2 signal, P3 signal) obtained from the corresponding adder 114 of each of the -3, and the component signal (C1 signal of the original color space) ) Is generated and stored in the image memory 116.

  The decorrelation conversion unit 117 responds to the component signal (C1 signal) of the original color space from the image memory 116 in response to a request for space or time prediction from the intra prediction / inter prediction unit 118. The decoded signals (C2 signal and C3 signal) of the other component signals of the original color space obtained from the corresponding image memories 116 of the compression encoding units 103-2 and 103-3 are input. The decorrelation obtained from the decorrelation parameters used for the decorrelation component signal at the time of encoding this time with respect to the decoded signal corresponding to the other component signal of the original color space at the time of the previous coding A reference signal for generating a prediction signal of a decorrelation component signal at the time of encoding by aligning the signal system by performing decorrelation conversion by a transformation matrix Generated and sent to the intra prediction / inter prediction unit 118.

  The intra-screen prediction / inter-screen prediction unit 118 reads the reference signal from the decorrelation conversion unit 117, performs intra-screen prediction or inter-screen prediction processing in the same manner as conventional MPEG and the like, and encodes this time. A prediction signal of the decorrelation component signal is generated and sent to the subtraction unit 111, and each parameter (P1 intra prediction parameter or motion vector) used for intra prediction or inter prediction processing is sent to the outside.

  The flow of generating and encoding a residual signal (P1 residual signal) using the prediction signal obtained from the intra-screen prediction / inter-screen prediction unit 118 will be described with reference to FIG. FIG. 4 schematically shows an example of intra-frame prediction / inter-screen prediction processing using a decorrelation component signal.

  The decorrelation conversion unit 101 receives a plurality of component signals (C1 signal, C2 signal, C3 signal) in the original color space, performs decorrelation conversion, and generates decorrelation component signals (P1 signal, P2 signal, (P3 signal) is generated (step S11). Subsequently, the decorrelation inverse transform unit 115 restores the decorrelation component signal (P1 signal) and the corresponding decorrelation component signals (P2 signal) obtained from the compression coding units 103-2 and 103-3. , P3 signal) is subjected to inverse transformation using the decorrelation inverse transformation matrix (step S12), and further, the decorrelation transformation unit 117 uses the decorrelation transformation matrix used for encoding this time to perform intra prediction / screening. As a reference signal used by the inter prediction unit 118, a new decorrelated component signal (Q1 signal: decorrelated component signal at the time of current coding, Q2: decorrelated component signal at the time of current coding, Q3: at the time of current coding Is generated (step S13).

  In the case of inter-screen prediction, the decorrelation conversion unit 101 uses the decorrelation matrix at the time of the current encoding to decorrelate the component signal (Q1 signal: the decorrelation component signal at the time of the current encoding, Q2: By generating a decorrelation component signal at the time of encoding this time, Q3: a decorrelation component signal at the time of encoding this time (step S14), the intra-screen prediction / inter-screen prediction unit 118 generated at step S13 A prediction signal can be generated on the same signal axis (see FIG. 3) using the reference signal (step S15).

  Thus, in order to generate and encode a residual signal (P1 residual signal) from the prediction signal obtained from the intra-screen prediction / inter-screen prediction unit 118, the reference used in the intra-screen prediction / inter-screen prediction unit 118 The residual signal is generated by making the decorrelation transformation matrix of the decorrelation component signal used when encoding the signal (when generating the prediction signal) the same. That is, in generating the reference signal, the encoded signal is locally decoded to generate a component signal before encoding, and then subjected to decorrelation inverse transformation by the decorrelation transformation matrix used at the time of encoding. The stored component signal is read out, and the reference signal is generated by performing the decorrelation conversion by the decorrelation conversion matrix used for the decorrelation component signal used for the current encoding. Therefore, when generating a prediction signal by predicting a signal in the time or space direction, it is preferable to convert the reference signal and the prediction signal so as to have the same signal axis. As a result, even when configured for performing decorrelation conversion, intra prediction or inter prediction used in MPEG or the like can be applied, and the amount of information to be transmitted can be further reduced. In this way, when decorrelation of the color space is performed not for the entire (moving) image but for each frame, slice, or macroblock, the reference signal used for intra prediction or inter prediction and the current prediction The signal system used as the reference signal is matched with the signal system that is currently subjected to the prediction process so that the three-axis signal system of the signal to be processed does not differ.

  In spatio-temporal signal prediction (intra-screen prediction or inter-screen prediction), a specific number of codes other than sequentially calculating a decorrelation transformation matrix for a decorrelation component signal for each coding block or each frame. The decorrelation transformation matrix for the decorrelation component signal may be sequentially calculated for each block or a specific number of frames. In particular, when a decorrelation transformation matrix common in one frame is used for intra-frame prediction, or when a decorrelation transformation matrix common to multiple frames is used for inter-frame prediction, encoding is performed with a reference signal. By making the decorrelation transformation matrixes of the decorrelation component signals used at the same time the processing load can be reduced. That is, the decorrelation conversion may be performed on the entire screen of one frame, or may be performed on a block basis including a plurality of pixels.

  According to the encoding device 10 of the first embodiment, the compression rate can be increased while suppressing loss to the overall image quality by performing the encoding after decorrelating. This method of reducing the amount of signal not only reduces the number of samples of the decorrelated component signal on the low contribution axis or coarsens the quantization but also the signal depth of the decorrelated component signal on the low contribution axis. Etc. On the other hand, the axis having a high contribution ratio can be configured to widen the dynamic range by performing fine dense step quantization.

  In the decorrelation transform, each of the original component signals at the corresponding pixel position is input in parallel, and for each combination of the original component signals, a principal component analysis is performed for each combination to obtain a decorrelation transform matrix. In the description, the example in which the decorrelation transformation matrix is encoded as information associated with the combination of the original component signals has been described. On the other hand, as another method for encoding, the closest one to the decorrelation transform matrix obtained by the principal component analysis method is selected from a finite number of decorrelation transform matrices, and the selected decorrelation transform matrix is selected. Is determined and encoded as auxiliary information associated with the combination of the original component signals. In this case, even in a decoding device that performs decorrelation inverse transformation, a finite number of decorrelation transformation matrices are defined in advance as decorrelation transformation matrices that can be identified by unique numbers in advance. In the processing, the uniqueness of the decorrelation transform matrix determined by selecting from the finite number of decorrelation transform matrices closest to the decorrelation transform matrix obtained by principal component analysis on the encoding side. A decorrelation transformation matrix is specified according to auxiliary information indicating a number, and decorrelation inverse transformation is executed. Thereby, the amount of information to be transmitted can be greatly reduced.

  Next, a decoding device that receives and decodes the bitstream encoded by the encoding device 10 according to the first embodiment will be described.

  FIG. 2 is a block diagram of the decoding apparatus according to the first embodiment of the present invention. The decoding device 20 of the present embodiment includes decoding units 201-1, 201-2, and 202-3 configured in the same manner, and decoding conversion units 202-1, 202-2, and 202-3 configured in the same manner. And a decorrelation inverse transform unit 203. The decoding unit 201-1 and the decoding conversion unit 202-1 are signal systems that receive and decode the P1 residual signal, and the decoding unit 201-2 and the decoding conversion unit 202-2 receive the P2 residual signal. The decoding unit 201-3 and the decoding conversion unit 202-3 are signal systems that receive and decode the P3 residual signal. The decoding conversion unit 202-1 includes an addition unit 211, an image memory 213, and an intra-screen prediction / inter-screen prediction unit 214.

  The decoding unit 201-1 receives and decodes the P1 residual signal encoded by the encoding unit 112 and sends it to the adding unit 211.

  The adding unit 211 adds the prediction signal obtained from the intra-screen prediction / inter-screen prediction unit 214 and the decoded signal of the P1 residual signal decoded by the decoding unit 201-1, and thereby decorrelates the component signal (P1 signal). Is sent to the decorrelation inverse transformation unit 203.

  The decorrelation inverse transform unit 203 generates the decorrelation component signal input from the decoding transform units 202-1, 202-2, and 202-3 from the decorrelation transform parameter obtained from the encoding device 10. Inverse transformation is performed using the decorrelation inverse transformation matrix to restore the original color space component signals (C1 signal, C2 signal, C3 signal).

  The intra-screen prediction / inter-screen prediction unit 214 reads the reference signal from the image memory 213 that stores the component signals (C1 signal, C2 signal, and C3 signal) of the original color space that has been inversely transformed, and outputs the reference signal from the encoding device 10. A prediction signal is generated using the obtained P1 intra prediction parameter / motion vector and sent to the adding unit 211.

  In this way, the decoding device 20 can reproduce the component signal in the original color space by decoding the decorrelated component signal.

  In addition, after decorrelating the color space on the encoding side, the signal quantity of the decorrelated component signal whose signal distribution is smaller than the total signal quantity among the decorrelated component signals is converted to another decorrelation. Since the encoding is performed by reducing the signal amount more than the signal amount of the component signal, the decoding device 20 can restore a signal with an increased compression rate while suppressing a loss in the overall image quality.

  Next, an encoding device and a decoding device according to Embodiment 2 of the present invention will be described.

  FIG. 5 is a block diagram of the encoding apparatus 30 according to the second embodiment of the present invention. The encoding apparatus 30 according to the second embodiment of the present embodiment inputs a plurality of component signals (C1 signal, C2 signal, C3 signal) in the original color space, and individually generates a residual signal. -1, 301-2, 301-3, decorrelation conversion section 302 for performing decorrelation conversion, sample number reduction / coarse quantization sections 303-1, 303-2, 303-3, and encoding section 304-1, 304-2, 304-3.

  The compression encoding units 301-1, 301-2, and 301-3 are configured in the same manner. The compression encoding unit 301-1 includes a subtraction unit 311, an encoding unit 312, a local decoding unit 313, and an addition. Unit 314, image memory 315, and intra-screen prediction / inter-screen prediction unit 316. The compression encoding unit 301-1 is configured in the same manner as an encoding device such as MPEG, and includes a subtraction unit 311, an encoding unit 312, a local decoding unit 313, an addition unit 314, and an intra-screen prediction / inter-screen prediction unit. The operations of 316 are the same as those of the subtracting unit 111, the encoding unit 112, the local decoding unit 113, the adding unit 114, and the intra prediction / inter prediction unit 118 in the first embodiment, respectively. Omitted. The intra-screen prediction / inter-screen prediction unit 316 in each of the compression encoding units 301-1, 301-2, and 301-3 reads a reference signal (C1 ′ signal) from the image memory 315 to generate a prediction signal. In addition to sending to the subtracting unit 311, the in-screen prediction parameter / motion vector is sent to the outside.

  The compression encoding unit 301-1 receives a component signal (C1 signal) of the original color space, and a residual signal (C1 ′ signal) obtained through the subtraction unit 311, the encoding unit 312, and the local decoding unit 313. Is sent to the adder 314 and sent to the decorrelation converter 302. Similarly, the compression encoding units 301-2 and 301-3 also receive the original color space component signals (C2 signal and C3 signal) and decorrelate the residual signals (ΔC2 ′ signal and ΔC3 ′ signal). To the conversion unit 302.

  The decorrelation conversion unit 302 applies the first embodiment to the residual signals (ΔC1 ′ signal, ΔC2 ′ signal, ΔC3 ′ signal) supplied from the compression encoding units 301-1, 301-2, and 301-3. The decorrelation conversion unit 101 performs the decorrelation conversion to generate the P1 residual signal, the P2 residual signal, and the P3 residual signal of the decorrelation component signal, and reduces the number of samples / coarse quantization, respectively. The data are sent to the sections 303-1, 303-2, and 303-3. The decorrelation conversion unit 302 generates a decorrelation component signal (P1 residual signal, P2 residual signal, P3 residual signal) of the residual signal, and generates a function constituting the decorrelation conversion matrix. It is sent to the outside as a decorrelation conversion parameter.

  The sample number reduction / coarse quantization unit 303-1 receives one of the decorrelation component signals obtained by decorrelating the component signals in the original color space from the decorrelation conversion unit 302, and determines the predetermined number. The signal amount is reduced according to the reduction rate with respect to the entire signal amount, and is sent to the encoding unit 304-1. In this example, it is assumed that signals are assigned as a P1 residual signal, a P2 residual signal, and a P3 residual signal in order from a signal with a low contribution ratio and a low contribution ratio. Therefore, the P1 residual signal is a decorrelation component signal of the residual signal of the axis with the lowest contribution of the signal distribution and having a low contribution rate, and the amount of this signal is used as the decorrelation component signal of the other residual signals. The signal amount is greatly reduced and transmitted to the encoding unit 304-1. This method of reducing the amount of signal reduces the number of samples of the decorrelated component signal on the axis with a low contribution rate or coarsens the quantization. As a result, the compression rate can be increased while suppressing loss in the overall image quality. The operations of the other sample number reduction / coarse quantization units 303-2 and 303-2 are the same.

  Each of encoding sections 304-1, 304-2, and 304-3 is a decorrelation component signal of the residual signal obtained from each of the sample number reduction / coarse quantization sections 303-1, 303-2, and 303-3. (P1 residual signal, P2 residual signal, P3 residual signal) is subjected to an encoding process according to a predetermined encoding procedure to generate a bitstream obtained by encoding a decorrelated component signal of the residual signal. Send it out. The encoding process can be configured to perform orthogonal transform, quantization, and variable length encoding, for example, as in MPEG.

  As described above, the second embodiment is configured to perform decorrelation conversion on the residual signal of the component signal in the original color space, appropriately reduce the signal amount, and send the signal to the outside. Therefore, the encoding apparatus 30 according to the second embodiment can apply the decorrelation conversion unit 302 with a relatively simple configuration to an existing encoding apparatus (conventional), and the advantages of the first embodiment. Can be included. Note that the compression encoding units 301-1, 301-2, and 301-3 may generate residual signals (ΔC1 ′ signal, ΔC2 ′ signal, and ΔC3 ′ signal) of a plurality of component signals in the original color space. Therefore, the encoding unit 312 and the local decoding unit 313 may be bypassed.

  Next, a decoding device 40 that receives and decodes the bitstream encoded by the encoding device 30 according to the second embodiment will be described.

  FIG. 6 is a block diagram of the decoding apparatus according to the second embodiment of the present invention. The decoding device 40 according to the present embodiment includes decoding units 401-1, 401-2, and 401-3, a decorrelation inverse conversion unit 402, and decoding conversion units 403-1, 403-2, and 403-3. .

  Each of decoding sections 401-1, 401-2, and 401-3 decodes a decorrelated component signal of the residual signal encoded by each of encoding sections 304-1, 304-2, and 304-3. To the decorrelation inverse transform unit 402.

  The decorrelation inverse transform unit 402 performs decorrelation transform obtained from the encoding device 30 on the decorrelation component signals of the residual signals input from the decoding units 401-1, 401-2, and 401-3. Inverse transformation is performed using the decorrelation inverse transformation matrix generated from the parameters to restore the residual signals (ΔC1 ′ signal, ΔC2 ′ signal, ΔC3 ′ signal) of the component signals in the original color space, and decode them respectively. The data is sent to the conversion units 403-1, 403-2, and 403-3.

  Each of the decoding conversion units 403-1, 403-2, and 403-3 is configured in the same manner, operates in the same manner as an existing decoding device such as MPEG (conventional), and includes an adding unit 411, an image memory 412, and An intra-screen prediction / inter-screen prediction unit 413 is provided.

  That is, the addition unit 411 includes the prediction signal obtained from the intra-screen prediction / inter-screen prediction unit 413 and the residual signal (ΔC1 ′ signal, ΔC2 ′ signal, ΔC3 ′ signal) obtained from the decorrelation inverse transformation unit 402. To restore the original color space component signal (C1 signal).

  Note that the intra-screen prediction / inter-screen prediction unit 413 reads the reference signal from the image memory 412 storing the restored original color space component signal (C1 signal), and obtains the C1 intra-screen prediction parameter obtained from the encoding device 30. / Generate a prediction signal using the motion vector and send it to the adder 411.

  In this way, the decoding device 40 can decode the decorrelated component signal. Therefore, the decoding device 40 of the second embodiment can apply the decorrelation inverse transform unit 402 with a relatively simple configuration to an existing (conventional) decoding device, and has the advantages of the first embodiment. All can be included.

  In other words, after decorrelating the color space on the encoding side, the signal quantity of the decorrelated component signal whose signal distribution is smaller than the total signal quantity among the decorrelated component signals is converted to another decorrelation. Since the encoding is performed by reducing the signal amount more than the signal amount of the component signal, the decoding device 40 can restore a signal with an increased compression rate while suppressing a loss in the overall image quality.

  Next, an encoding device and a decoding device according to Embodiment 3 of the present invention will be described.

  FIG. 7 is a block diagram of the encoding apparatus 50 according to the third embodiment of the present invention. The encoding apparatus 50 according to the third embodiment of the present embodiment inputs a plurality of component signals (C1 signal, C2 signal, C3 signal) in the original color space, individually generates a residual signal, and performs orthogonal transformation. Further, compression encoders 502-1, 502-2, and 502-3 for generating a bitstream by quantizing the decorrelation component signal of the orthogonal transform coefficient obtained through the decorrelation transformer 501; A decorrelation conversion unit 501 that performs correlation conversion.

  The compression encoding units 502-1, 502-2, and 502-3 are configured in the same manner. The compression encoding unit 502-1 includes a subtraction unit 511, an orthogonal transform unit 512, a quantization unit 513, and an inverse. A quantization unit 514, a decorrelation inverse transformation unit 515, an inverse orthogonal transformation unit 516, an addition unit 517, an image memory 518, and an intra-screen prediction / inter-screen prediction unit 519 are provided. Also, the subtraction unit 511, orthogonal transform unit 512, quantization unit 513, inverse quantization unit 514, inverse orthogonal transform unit 516, addition unit 517, and intra prediction / inter prediction unit 519 of the compression encoding unit 502-1. These operations are the same as those performed by an encoding device such as MPEG. The intra-screen prediction / inter-screen prediction unit 519 in each of the compression encoding units 502-1, 502-2, and 502-3 reads a reference signal (C1 ′ signal) from the image memory 518 and generates a prediction signal. While sending to the subtraction part 511, the prediction parameter / motion vector in a screen is sent outside.

  The subtraction unit 511 receives the component signal (C1 signal) of the original color space, and sends a residual signal for the prediction signal obtained from the intra prediction / inter prediction unit 519 to the orthogonal transform unit 512.

  The orthogonal transform unit 512 performs orthogonal transform (for example, DCT) on the residual signal obtained from the subtracting unit 511, and the signal (C1_D signal) of the orthogonal transform coefficient of the residual signal is sent to the decorrelation transform unit 501. Send it out.

  The decorrelation transform unit 501 receives signals of orthogonal transform coefficients (C1_D signal, C2_D signal, C3_D signal) of residual signals generated by the compression coding units 502-1, 502-2, and 502-3. Then, the decorrelation transform described in the first embodiment is executed to generate decorrelation component signals (P1 signal, P2 signal, P3 signal) of orthogonal transform coefficients, respectively, and the corresponding compression encoding units 502- 1, 502-2, and 502-3 are sent to the quantization unit 513. The decorrelation conversion unit 501 sends the information of the decorrelation conversion matrix used for the decorrelation conversion to the outside as a decorrelation conversion parameter and stores it in the decorrelation parameter storage unit 520.

  The quantization unit 513 performs a quantization process similar to MPEG or the like on the decorrelation component signal (P1 signal) of the orthogonal transform coefficient, generates a P1 residual signal stream, and sends the P1 residual signal stream to the outside. The data is sent to the quantization unit 514.

  The inverse quantization unit 514 performs inverse quantization on the P1 residual signal supplied from the quantization unit 513 and transmits the result to the decorrelation inverse transform unit 515.

  As in the case of the first embodiment, the decorrelation inverse transformation unit 515 is configured to obtain a decorrelation transformation matrix used for the decorrelation component signal at the time of the previous encoding. A decorrelation parameter is read from 520, a decorrelation inverse transform matrix is generated from the decorrelation parameter, and a decompressed signal (P1 signal) of the decorrelation component signal obtained from the inverse quantization unit 514, and a compression code The decorrelation component signals (P2 signal, P3 signal) obtained from the corresponding inverse quantization units 514 of the quantization units 502-2 and 502-3 are subjected to inverse transformation using a decorrelation inverse transformation matrix, and orthogonal A decoded signal corresponding to the residual signal (C1_D signal) of the transform coefficient is generated and sent to the inverse orthogonal transform unit 516.

  The inverse orthogonal transform unit 516 performs inverse orthogonal transform on the decoded signal corresponding to the residual signal (C1_D signal) of the orthogonal transform coefficient obtained from the decorrelation inverse transform unit 515, and performs component signal conversion in the original color space. A residual signal is generated and sent to the adder 517.

  The adder 517 adds the residual signal of the component signal in the original color space obtained from the inverse orthogonal transform unit 516 and the prediction signal obtained from the intra prediction / inter prediction unit 519 to obtain the next encoding. A restoration signal for the reference signal used in the above is generated and stored in the image memory 518.

  The intra-screen prediction / inter-screen prediction unit 519 reads the reference signal (C1 ′ signal) stored in the image memory 518, and performs intra-screen prediction or inter-screen prediction processing in the same manner as in the conventional MPEG or the like. A prediction signal of a decorrelated component signal at the time of encoding is generated and sent to the subtraction unit 511 and the addition unit 517, and each parameter (C1 intra prediction parameter or motion used for intra prediction or inter prediction processing) is used. Vector) to the outside.

  Note that the output of the quantization unit 513 can be configured to be subjected to variable length coding, for example, in the same way as MPEG.

  In the third embodiment, the quantization unit 513 can implement the sample number reduction / coarse quantization processing described in the first or second embodiment.

  As described above, the third embodiment is configured to perform orthogonal transform on the residual signal of the component signal in the original color space, further perform decorrelation transform, and then quantize and send the signal to the outside. ing. Therefore, the coding apparatus 50 according to the third embodiment can apply the decorrelation conversion unit 501 with a relatively simple configuration to an existing coding apparatus (conventional), and the advantages of the first embodiment. Can be included.

  Next, a decoding device 60 that receives and decodes the bitstream encoded by the encoding device 50 according to the third embodiment will be described.

  FIG. 8 is a block diagram of the decoding apparatus according to the third embodiment of the present invention. The decoding device 60 according to the present embodiment includes an inverse quantization unit 601-1, 601-2, and 601-3, a decorrelation inverse transform unit 602, and an inverse orthogonal transform unit 603-1, 603-2, and 603-3. And decoding conversion units 604-1, 604-2, and 604-3.

  Each of the inverse quantization units 601-1, 601-2, and 601-3 is a decorrelation component signal of the residual signal encoded through each of the decoding conversion units 604-1, 604-2, and 604-3. Is sent to the decorrelation inverse transformation unit 602.

  The decorrelation inverse transform unit 602 decorrelates the decorrelation component signals of the residual signals input from the inverse quantization units 601-1, 601-2, and 601-3, and obtains the decorrelation obtained from the encoding device 50. Inverse transformation is performed using the decorrelation inverse transformation matrix generated from the generalization transformation parameters to restore the residual signal of the orthogonal transformation coefficient for the component signal in the original color space, and the inverse orthogonal transformation units 603-1, 603-1, respectively. It is sent to 603-2 and 603-3.

  Each of the inverse orthogonal transform units 603-1, 603-2, and 603-3 performs inverse orthogonal transform on the residual signal of the orthogonal transform coefficient obtained from the decorrelation inverse transform unit 602, and the original color space A residual signal of the component signal is generated and sent to each of the decoding conversion units 603-1, 603-2, and 603-3.

  Each of the decoding conversion units 603-1, 603-2, and 603-3 is configured in the same manner, operates in the same manner as an existing decoding device (conventional) such as MPEG, an addition unit 611, an image memory 612, An intra-screen prediction / inter-screen prediction unit 613 is provided.

  That is, the adding unit 611 adds the prediction signal obtained from the intra-screen prediction / inter-screen prediction unit 613 and the residual signal of the component signal in the original color space obtained from the decoding conversion unit 603-1 to obtain the original signal. The color space component signal (C1 signal) is restored.

  The intra prediction / inter prediction unit 613 reads the reference signal from the image memory 612 storing the restored original color space component signal (C1 signal), and obtains the C1 intra prediction parameter obtained from the encoding device 50. / Generate a prediction signal using the motion vector and send it to the adder 611.

  In this way, the decoding device 60 can decode the decorrelated component signal. Therefore, the decoding device 60 according to the third embodiment can apply the decorrelation inverse transform unit 602 with a relatively simple configuration to an existing (conventional) decoding device, and has the advantages of the first embodiment. All can be included.

  In other words, after decorrelating the color space on the encoding side, the signal quantity of the decorrelated component signal whose signal distribution is smaller than the total signal quantity among the decorrelated component signals is converted to another decorrelation. Since the encoding is performed by reducing the signal amount more than the signal amount of the component signal, the decoding device 60 can restore a signal with an increased compression rate while suppressing a loss in the overall image quality.

  Therefore, in any of the embodiments, according to the present invention, the entire signal amount of the uncorrelated component signals is utilized by utilizing the correlation between the component signals that has not been sufficiently used in the conventional image coding scheme. On the other hand, since the signal amount of the decorrelated component signal having a small signal distribution is reduced, further compression is possible as compared with the conventional image coding method.

  Furthermore, as one aspect of the present invention, each of the encoding device and the decoding device of each embodiment can be configured as a computer. A program for causing a computer to realize each component in each of the encoding device and the decoding device of each of the above-described embodiments is stored in a storage unit provided inside or outside the computer. Such a storage unit can be realized by an external storage device such as an external hard disk or an internal storage device such as ROM or RAM. The control unit provided in the computer can be realized by controlling a central processing unit (CPU) or the like. In other words, the CPU can appropriately read from the storage unit a program in which the processing content for realizing the function of each component is described, and realize the function of each component on the computer. Here, the function of each component may be realized by a part of hardware.

  For example, FIG. 9 shows a control flow in a computer configured as an encoding apparatus according to an embodiment of the present invention. Referring to FIG. 9, a signal element (pixel value or orthogonal transform coefficient) of each component signal is input to a computer that is configured as an encoding device, and decorrelation conversion is performed (step S1). The encoded signal is encoded (step S2), and the encoded bitstream and the information of the decorrelation matrix used for the decorrelation conversion are sent to the outside as decorrelation parameters ( Step S3) It can be realized as a program for executing the process.

  For example, FIG. 10 shows a control flow in a computer configured as a decoding apparatus according to an embodiment of the present invention. Referring to FIG. 10, a computer configured as a decoding device includes a bitstream that has been subjected to encoding processing, and a decorrelation parameter that includes information on a decorrelation matrix used for decorrelation conversion on the encoding device side. (Step S21), the encoded bitstream is decoded (step S22), the decorrelation inverse transformation is performed using the decorrelation parameter, and the signal element of the component signal in the original color space (Step S23) can be realized as a program for executing the process.

  In addition, the program describing the processing contents can be distributed by selling, transferring, or lending a portable recording medium such as a DVD or CD-ROM, and such a program can be distributed on a server on a network, for example. Can be distributed by transferring the program from the server to another computer via the network.

  In addition, a computer that executes such a program can temporarily store, for example, a program recorded on a portable recording medium or a program transferred from a server in its own storage unit. As another embodiment of the program, the computer may directly read the program from a portable recording medium and execute processing according to the program, and each time the program is transferred from the server to the computer. In addition, the processing according to the received program may be executed sequentially.

  While the embodiments of the present invention have been described in detail with specific examples, it will be apparent to those skilled in the art that various modifications and changes can be made without departing from the scope of the claims of the present invention.

  According to the present invention, since further compression is possible as compared with the conventional image encoding method, it is useful for the application of an encoding technique for encoding a component signal in a color space.

DESCRIPTION OF SYMBOLS 10 Encoding apparatus 20 Decoding apparatus 30 Encoding apparatus 40 Decoding apparatus 50 Encoding apparatus 60 Decoding apparatus 101 Correlation conversion part 103-1,103-2,103-3 Compression encoding part 110 Sample number reduction / coarse quantization Unit 111 subtracting unit 112 encoding unit 113 local decoding unit 114 addition unit 115 decorrelation inverse transformation unit 116 image memory 117 decorrelation transformation unit 118 intra prediction / inter prediction unit 119 decorrelation parameter storage unit 201-1 , 201-2, 202-3 decoding units 202-1, 202-2, 202-3 decoding conversion unit 203 decorrelation inverse conversion unit 211 addition unit 213 image memory 214 intra prediction / inter prediction unit 301-1, 301-2, 301-3 Compression encoding unit 302 Decorrelation conversion unit 303-1, 303-2, 303-3 for performing decorrelation conversion Reduction / coarse of sample Child units 304-1, 304-2, 304-3 Encoding unit 311 Subtraction unit 312 Encoding unit 313 Local decoding unit 314 Addition unit 315 Image memory 316 Intra-screen prediction / inter-screen prediction units 401-1, 401-2 , 401-3 decoding unit 402 decorrelation inverse transformation unit 403-1, 403-2, 403-3 decoding transformation unit 411 addition unit 412 image memory 413 intra prediction / inter prediction unit 502-1, 502-2, 502-3 Compression encoding unit 501 Decorrelation transform unit 511 Subtraction unit 512 Orthogonal transform unit 513 Quantization unit 514 Inverse quantization unit 515 Decorrelation inverse transform unit 516 Inverse orthogonal transform unit 517 Addition unit 518 Image memory 519 In screen Prediction / inter-screen prediction unit 601-1, 601-2, 601-3 inverse quantization unit 602 decorrelation inverse transform unit 603-1, 603-2, 603-3 inverse orthogonal transform unit 04-1,604-2,604-3 decoded transform unit 611 adding unit 612 image memory 613 intra prediction / inter prediction section

Claims (13)

An encoding device that compresses an image by converting a color space formed by a plurality of component signals into an uncorrelated color space,
Decorrelation conversion means for inputting a signal element of each component signal in a predetermined color space and executing decorrelation conversion;
An encoding means for generating an encoded signal by performing an encoding process on the decorrelated signal;
The decorrelation conversion unit generates information of a decorrelation matrix used for the decorrelation conversion as a decorrelation parameter.   The decorrelation conversion means determines principal components in descending order of eigenvalues using a principal component analysis method, and converts the signal elements of each component signal into signal elements of the decorrelated component signal composed of the principal components. The encoding apparatus according to claim 1, wherein a decorrelation transformation matrix is configured to generate the decorrelation parameters.   The signal amount is reduced in accordance with a predetermined reduction rate with respect to the entire signal amount with respect to one or more decorrelation component signals among the decorrelation component signals immediately after being decorrelated by the decorrelation conversion unit. The encoding apparatus according to claim 1, further comprising a signal amount reduction unit.   The signal amount reducing means reduces the signal amount by reducing the number of samples with respect to the decorrelated component signal having the smallest signal distribution among the decorrelated component signals immediately after the decorrelation by the decorrelation converting means. The encoding device according to claim 3, wherein the encoding device is reduced.   The signal amount reduction means performs coarse quantization on a decorrelation component signal having the smallest signal distribution among the decorrelation component signals immediately after decorrelation by the decorrelation conversion means, thereby reducing the signal quantity. The encoding apparatus according to claim 3, wherein the encoding apparatus is reduced. Signal prediction means for generating a prediction signal by predicting a signal in a temporal or spatial direction;
A signal axis conversion means for converting the reference signal used for the signal prediction means and the prediction signal to have the same signal axis;
The encoding device according to claim 1, further comprising: A finite number of decorrelation transformation matrices are defined in advance as a decorrelation transformation matrix that can be identified by a unique number in advance.
The decorrelation transforming means selects from the finite number of decorrelation transform matrices closest to the decorrelation transform matrix obtained by principal component analysis, and the unique number of the selected decorrelation transform matrix Is determined as auxiliary information associated with the combination of the original component signals. 7. The encoding apparatus according to claim 1, wherein A decoding device that decodes an image-compressed signal by converting a color space formed by a plurality of component signals into an uncorrelated color space,
A decorrelation conversion is performed on the signal elements of each component signal in a predetermined color space, and the encoded signal subjected to the encoding process and the decorrelation matrix information used for the decorrelation conversion are not included. Decoding means for obtaining a correlation parameter and decoding the encoded signal;
Decorrelation inverse transform means for performing decorrelation inverse transform using the decorrelation parameter and restoring signal elements of the component signal in the original color space;
A decoding apparatus comprising:   The decorrelation inverse transformation means constructs a decorrelation inverse transformation matrix that inversely transforms the signal element of the decorrelation component signal into the signal element of each component signal in the original color space using the decorrelation parameter. The decoding device according to claim 8, wherein: Signal prediction means for generating a prediction signal by predicting a signal in a temporal or spatial direction;
A signal axis conversion means for converting the reference signal used for the signal prediction means and the prediction signal to have the same signal axis;
The decoding device according to claim 9, further comprising: A finite number of decorrelation transformation matrices are defined in advance as a decorrelation transformation matrix that can be identified by a unique number in advance.
The decorrelation inverse transform means is determined by selecting the closest one from the finite number of decorrelation transform matrices to the decorrelation transform matrix obtained by principal component analysis on the encoding side, The decoding device according to claim 9 or 10, wherein a decorrelation transformation matrix is specified according to auxiliary information indicating a unique number of the decorrelation transformation matrix, and the decorrelation inverse transformation is executed. In a computer configured as an encoding device that compresses an image by converting a color space formed by a plurality of component signals into an uncorrelated color space,
Inputting a signal element of each component signal in a predetermined color space and performing a decorrelation conversion;
Performing a coding process on the decorrelated signal to generate a coded signal;
Generating information of a decorrelation matrix used for the decorrelation transform as a decorrelation parameter;
A program for running A computer configured as a decoding device that decodes an image-compressed signal by converting a color space formed by a plurality of component signals into an uncorrelated color space;
A decorrelation conversion is performed on the signal elements of each component signal in a predetermined color space, and the encoded signal subjected to the encoding process and the decorrelation matrix information used for the decorrelation conversion are not included. Obtaining a correlation parameter;
Decoding the encoded signal;
Performing a decorrelation inverse transform using the decorrelation parameters to recover the signal elements of the component signal in the original color space;
A program for running

Images