JPS63190483A - Variable sampling transmission system - Google Patents

Abstract

PURPOSE:To attain efficient signal visually by using a inverted phase sampling means in scanning line sequential as a smapling pattern in the mode of a maximum density sampling picture element. CONSTITUTION:In interpolating a non-sample point from a pattern by an inter- field offset sub sampling to a still picture, the interpolation between fields is directed unidirectionally to cause an disadvantage. As the countermeasures, in deviating the position of a still picture by 1/2 line in the opposite direction in advance by an encoder, it results in avoiding the movement of the still picture position. The 1/2 line movement of the picture position, that is, is inverted vertically at each frame. A broken line field signal moved upward by 1/2 line is moved upward by 1/2 line at the decoder and interpolated into the solid line field signal to reproduce the still picture signal of the solid line field. Similarly, the solid line field signal moved downward by 1/2 line is interpolated to the broken line field signal moved downward by 1/2 line to reproduce the still picture signal of broken line field.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a band compression transmission system for high-definition television signals, and particularly relates to an improvement of a TAT (time axis conversion) band compression transmission system.

(Prior Art) High-definition television signals have a large amount of information to be transmitted, and band compression transmission while maintaining high image quality is an essential element for the development of high-definition television systems.

One of the band compression transmission methods is the TAT (time axis conversion) band compression transmission method. An outline of the TAT method will be described below to facilitate understanding of this invention, but details can be found in References (1) Tanimoto et al.: Bandwidth Compression of High-Definition Television Using the TAT (Time Axis Transformation) Method, Journal of the Society of Television Engineers, Vol. , 39, no.

10, pp. 84-90 (1985), see document (2) JP-A-61-62286: Image signal band compression method.

Bandwidth compression can be achieved by extracting some of the pixels that make up a television image and rearranging them at wider time intervals. At this time, there is no problem in thinning out pixels in areas that are not fine, and if frame memory is used, pixels can be significantly thinned out even in stationary areas or areas that can be effectively considered stationary. The TAT method takes advantage of this, first coarsely sampling the entire screen to extract pixels (basic pixels) that represent the basic structure of the image, and then selecting from the remaining pixels to represent the fineness depending on the nature of the image. Extract the pixels required for (additional pixels). If the total of basic pixels and additional pixels is made smaller than the original number of pixels, and these are arranged at uniform time intervals wider than the original pixel interval and transmitted as an analog signal, band compression transmission will be achieved.

FIG. 4(a) shows a flowchart showing the basic processing of the TAT method for the transmitting side and the receiving side, respectively.

(b), and the fourth
Shown in Figure (C). In FIG. 4(C), circles and squares represent sampling pixels of even and odd fields, respectively, * and ■ are basic pixels, and circles and old marks are additional pixels. Also, the X marks in the figure are non-sampled pixels, which are obtained by interpolation on the receiving side, and by providing the X marks, the amount of information to be transmitted is reduced to 1/2.

As the basic pixel, 1/4 of the total sampling pixels are taken as shown by the * mark and the cabinet seal in Figure 4 (C), and the band is further increased by 1.
The case of compression to /2 will be explained as an example.

On the transmitting side, the sampling pixels of the original image 11 are uniformly thinned out to 1/4 by thinning 12 to extract basic pixels. Next, additional pixels are extracted from the remaining pixels, and the mode decision 15 is used to determine whether or not to take additional pixels. This is done by checking the fineness of each image.

In order to perform mode determination 15, 3/4 pixels other than the basic pixels are interpolated by interpolation 13 from the basic pixels and the restored image signal of the previous frame stored in the frame memory. In other words, a one-frame difference is created for basic pixels at a common position in each frame, and based on that value, it is determined whether each block is a moving area or a still area. Perform interframe interpolation from pixels of the previous frame. The image obtained in this way is called a basic image, and the difference between this basic image and the original image (interpolation error) is determined by the difference 14 from the original image, and blocks with large interpolation errors are regarded as fine parts, and the blocks with large interpolation errors are sorted in descending order. 1 in order
The block /3 is a block to which an additional pixel is added by the additional pixel determination 16.

At this time, the total number of additional pixels becomes 1/4 of the total number of sample pixels, which together with 1/4 of the previous basic number of pixels becomes 1/2, and the amount of information to be transmitted is compressed to 1/2.

On the other hand, on the receiving side, pixels other than the basic pixels are reproduced from the basic pixels and the restored image signal of the previous frame by the same interpolation 17 as on the transmitting side to obtain a basic image. Next, in accordance with the mode information, the interpolated pixels are replaced with additional pixels to obtain a 18L fine reproduced image 19.

(Problems to be Solved by the Invention) The in-field sampling pattern of the TΔ lower band compression transmission system is shown again in FIG. 5(a). - A mark indicates a sampling position for a basic pixel that is always transmitted, and an O mark indicates a sampling position for an additional pixel that is additionally transmitted as required for high-density transmission.

If we consider the two-dimensional spatial frequency characteristics of the currently transmitted image with respect to this field transmission pattern, it will be as shown in Fig. 5, etc.). The horizontal axis of the figure is the horizontal spatial frequency, and the vertical axis is the vertical spatial frequency.The pattern shown in this figure is a transmission method that is wasteful to human vision, that is, it has a large resolution in diagonal directions where it is less necessary. I understand.

Therefore, an object of the present invention is to improve this visually wasteful transmission method.

(Means for Solving the Problems) That is, the variable sampling transmission method of the present invention is a band compression variable sampling transmission method in which an image is restored by transmitting an appropriate combination of additional pixels to basic pixels. The use of inverted phase sampling (line, quincunx) means in scanning line sequential order as a sampling pattern in the above is referred to as vF fine.

Further, in a preferred embodiment of the present invention, the inverted phase sampling means is a transmission-side encoder, and the vertical position of the sampling pixel after sampling of the odd field and the even field in the static region is alternately adjusted upward and downward by 172 degrees, respectively, for each frame. It is characterized in that it includes means for transmitting data by shifting each line.

(Examples) The present invention will be described in detail below by way of examples with reference to the accompanying drawings.

FIG. 1(a) shows an intra-field sampling pattern according to the method of the present invention, and FIG. 1(b) shows two-dimensional spatial frequency characteristics of an image transmitted by this sampling pattern. By comparing the characteristics shown in Fig. 1(b) with the characteristics shown in Fig. 5, etc., it can be seen that the characteristics shown in Fig. 1(b) are efficient transmission characteristics that also take into account the human visual system. .

However, it has been found that using only the sampling pattern described above causes problems in image restoration.This is because when trying to interpolate non-sample points from a pattern using inter-field offset sub-sampling for a still image, inter-field offset sub-sampling The ring pattern is shown in Figure 2 (
a) As shown in the pattern shown in the figure, for example, interfield interpolation using adjacent pixels of the non-sample point marked X is vertically asymmetrical, that is, in the case of this figure, the pixel signal that serves as the reference for interpolation is only in the upper part. , the so-called inter-field interpolation is
It becomes inconvenient to become ``one-handed.'' In the case of FIG. 2(a), of course, if the solid lines are the scanning lines of even fields, the broken lines are the scanning lines of odd fields.

As a way to solve the above-mentioned disadvantages, we focused on the fact that inter-field interpolation is performed only in the stationary area, and changed the pattern used for high-density transmission to the stationary area to the high-density transmission to the moving area. For example, a sampling pattern shown in FIG. 3 may be used instead of the pattern used when In the figure, non-sample points, such as the x marks, indicate that interfield interpolation is ``ambidextrous'' with respect to the top and bottom.

However, although this method achieves perfect inter-field interpolation, the sampling pattern differs between the moving region and the still region, making the configuration of the interpolation filter complex, and the phase of the sampling that is always being transmitted is different between moving and static regions. This makes it difficult to detect motion on the receiving side.

The present invention separately proposes an embodiment in which five of the sampling patterns shown in FIG. 2(a) are used, that is, the same sampling pattern is used with no difference in the moving area and the stationary area. In the case of the pattern shown in Figure 2(a), inter-field interpolation is the simplest example of interpolation from the previous field (dashed line) to the current field (solid line), as shown by the arrow in Figure 2(b). The interpolation is performed in a "one-handed" manner as explained in Figure 2 (a), and if the vertical position of the resulting still image is equivalently 1/2 line (Figure 2, etc.), it is shifted downward. That means that.

As a countermeasure against this, as shown in Figure 2 (C),
If the position of the still image is shifted in advance by 1/2 line in the opposite direction (upward in the case of FIG. 2(b)) on the encoder side, the position of the still image does not need to be moved as a result.

The 1/2 line movement direction of the image position is reversed upward and downward for each frame. On the decoder side, the broken line field signal moved upward by 1/2 line is interpolated into the solid line field signal moved upward by 1/2 line to reproduce the still image signal of the solid line field. Similarly, the solid line field signal shifted downward by 1/2 line is interpolated into the broken line field signal shifted downward by 1/2 line to reproduce the still image signal of the broken line field. If this is done, the vertical resolution of the high frequency component will decrease, but it is possible to perform inter-field interpolation.

(Effects of the Invention) By applying the method of the present invention to a band compression variable sampling transmission method that restores an image by transmitting an appropriate combination of basic pixels and additional pixels, visually efficient signal transmission can be achieved. Interfield interpolation can be performed for still images without any inconvenience, and signals with sufficient horizontal resolution can be transmitted even for moving images.

[Brief explanation of the drawing]

Fig. 1(a) shows the in-field sampling pattern according to the method of the present invention, Fig. 1(a) shows the two-dimensional spatial frequency characteristics that can be transmitted in Fig. 1(a), and Fig. 2(a) shows the in-field sampling pattern according to the method of the present invention. The intra-frame sampling pattern considering inter-field interpolation is shown in the same figure (b).
shows the interfield interpolation pattern from the dashed line field to the solid line field in the sampling pattern shown in (a) of the same figure, and (c) shows the 1/1 interpolation pattern resulting from the one-sided interpolation generated by the method of the present invention.
A schematic block diagram of the configuration of an embodiment for preventing two-line position shift is shown, FIG. 3 shows an example of a sampling pattern for a stationary area to prevent one-sided interpolation due to interfield interpolation, and FIG. 5 shows the basic processing flowchart of the conventional TAT method on the transmitting side (a), the receiving side (b), and its sampling pattern (C). A sampling pattern (a) and a two-dimensional spatial frequency characteristic (b) that can be transmitted by the sampling pattern are shown. 1... Move 1/2 line upward 2... Move 1/2 line downward 3... Broken line → solid line Interpolation between fields 4... Solid line → Broken line interpolation between fields 11... Original picture 12 ... Thinning 13.1
7...Interpolation 14...Difference from original image 15...Mode determination 16...Additional pixel determination 18...Replacement 19...Reproduced image patent applicant Date
Japan Broadcasting Association Figure 1 (a) Figure 1 (b) Horizontal 7 opening liquid fall - Figure 3

Claims (1)

[Claims] 1. In a band compression variable sampling transmission method that restores an image by transmitting an appropriate combination of basic pixels and additional pixels, as a sampling pattern in the mode of maximum density sampling pixels, scanning lines are sequentially inverted in phase. A variable sampling transmission method characterized by using sampling (line, quincunx) means. 2. The inversion phase sampling means is an encoder on the transmitting side, and alternately shifts the vertical position of the sampling pixel after sampling the odd field and even field in the stationary area by 1/2 line up and down for each frame. The variable sampling transmission system according to claim 1, further comprising means for transmitting.