JPH0785266A - Image rotating device - Google Patents

Abstract

PURPOSE:To attain high speed image rotating processing without magnifying an original picture in its vertical direction. CONSTITUTION:This picture rotating device is provided with a 1st memory 101 for storing an original picture, an input processing part 102 for inputting the image data of the original picture from the memory 101, a horizontally magnifying/reducing processing part 103 for converting the number of horizontal picture elements in the image data obtained from the processing part 102, an output processing part 104 for writing the image data obtained from the processing part 103 in a specified address of the 2nd memory 105, and an overall control part 106 for controlling the whole device. At the time of arranging respective picture elements in a result image obtained by rotating the original picture on respective grating points on a two-dimensional coordinate system, the processing part 104 calculates the number of picture elements of the result image existing on the same horizontal grating line and simultaneously writes the data of each word unit to be stored in the same word stored in the 2nd memory 105.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image rotating device for rotating a digital image.

[0002]

2. Description of the Related Art In recent years, with the progress of computer utilization technology, the number of processes for editing images on a computer is increasing. Especially, in order to speed up the image rotation process and improve the image quality, an image rotation device is required. It has come to be used. This image rotation device is known, for example, in the structure disclosed in Japanese Patent Laid-Open No. 63-191192. The conventional image rotation device will be described below.

The processing in the conventional image rotating apparatus is the number-of-pixels conversion processing for calculating a vertically enlarged and horizontally reduced image according to the rotation angle θ from the original image, and the grid point coordinates of the transfer destination along the straight line of the inclination θ. Pixel transfer processing for transferring the vertically enlarged and horizontally reduced image to the lattice points.

The rotation processing of a one-dimensional image in such a conventional image rotation apparatus will be described with reference to FIG. FIG. 11 is for explaining the processing in the case of rotating the one-dimensional original image by the angle θ with the center of rotation as the origin. In FIG. 11, the origin is passed and the inclination is tan
The straight line of θ is represented by L, and the intersections of this straight line L and each vertical grid line are represented by (E ₀ , E ₁ , E ₂ , ...) In order from the left. Also, (E
₀ , E ₁ , E ₂ , ...) is the closest lattice point to (C ₀ , C
₁ , C ₂ ...).

In the rotation processing of a one-dimensional image, the original image is reduced according to the rotation angle, and each pixel of the reduced image is sequentially
This can be realized by transferring to the lattice points (C ₀ , C ₁ , C ₂ ...) Calculated from the straight line L. Hereinafter, (C ₀ ,
Grid points for transferring pixels of the original image, such as C ₁ , C ₂ ...), are referred to as transfer grid points.

The procedure of processing a one-dimensional image composed of W pixels will be described below. Step 1.1 The x coordinate XC _i and the y coordinate YC _i of the closest grid point C _i to the intersection E _i (i is an integer) of the straight line L and the vertical grid line are calculated from the following formulas. However, at this time, i is 0 ≦ i <[Wco
sθ + 0.5] is satisfied. As a result, [Wc
The coordinates of the os θ + 0.5] grid points C _i are obtained. XC _i = i (1) YC _i = [i × tan θ + 0.5] (2) Note that [x] represents the maximum integer not exceeding x.

Step 1.2 The original image having W pixels is represented by [Wcos θ + 0.
5] The image is reduced to an image composed of pixels.

Step 1.3 Each pixel of the image reduced in the process of step 1.2 is sequentially transferred to the [Wcos θ + 0.5] lattice points calculated in step 1.1. In this way, the rotation processing of the one-dimensional image can be performed.

Next, the rotation processing of a two-dimensional rectangular image will be described with reference to FIG. FIG. 12 shows a case where a two-dimensional image is rotated by an angle θ with the center of rotation as the origin.
This is for explaining the conventional processing. In FIG. 12, it is assumed that the y-intercept is k (k is an integer) and the straight line having a slope tan θ is L _k . That is, the straight line L _k is represented as follows. L _k : y = (tan θ) x−k (3) Further, the inclination passes through the origin and the inclination is (−tan (π / 2−θ)).
Is represented by G. That is, the straight line G is expressed as follows. G: y =-(tan (π / 2−θ)) x (4)

To rotate a two-dimensional rectangular image, the original image is first enlarged in the vertical direction, and each pixel of the kth horizontal one line of the enlarged image is transferred to a transfer grid point calculated from the straight line L _k. It can be realized by transferring. However, at this time, the first pixel of the k-th horizontal 1-line of the enlarged image is transferred to the closest grid point from the intersection of the straight line L _k and the straight line G. In this specification, a grid point to which the first pixel of each horizontal 1 line is transferred is referred to as a transfer grid point. In the conventional image rotation device, the transfer grid points are straight lines L _k.
The grid point is the closest to the intersection of the line G and the line G.

Next, the procedure of the processing of the conventional image rotation apparatus when rotating a two-dimensional rectangular image having W horizontal pixels and H vertical pixels will be described. Step 2.1 A process of vertically enlarging the original image having H pixels in the vertical direction is performed so that the number of pixels in the vertical direction becomes [H / cos θ + 0.5].

Step 2.2 The above steps 1.1 to 1.3 are performed for each horizontal line of the two-dimensional rectangular image enlarged in the vertical direction in the step 2.1. . However, when processing the kth horizontal one line of the two-dimensional rectangular image enlarged in the vertical direction in the process of step 2.1, the straight line L shown in step 1.1 is changed to the straight line Lk, and the range of i Is processed as follows. This is because the x coordinate of the intersection of the straight line Lk and the straight line G is expressed as kcos θ sin θ. i: [kcos θ sin θ + 0.5] ≦ i <[k cos θ sin θ + 0. 5] + [Wsin θ + 0.5] (5) By performing the above processing, the rotation processing of the two-dimensional image can be performed.

Next, the structure of a conventional image rotation device based on the above principle will be described with reference to FIG. FIG. 13 shows a conventional image rotation device. FIG.
In the figure, 1301 is a memory for storing the original image. A rotation angle holding device 1302 holds the rotation angle information of the image. A pixel number conversion unit 1303 converts the number of pixels of an image by inputting the original image from the memory 1301 and reducing the original image in the horizontal direction and in the vertical direction based on the output from the rotation angle holding device 1302. It is a device. Reference numeral 1304 is a grid point coordinate calculation device that calculates the coordinates of the grid points based on the output from the rotation angle holding device 1302. An x-coordinate storage device 1305 stores the x-coordinates of the grid points calculated by the grid point coordinate calculation device 1304. 1306 is a y-coordinate storage device that stores the y-coordinates of the grid points calculated by the grid-point coordinate calculation device 1304.
x-coordinate storage device 1305 and y-coordinate storage device 1306
Is output to the grid point coordinate calculation device 1304 and is used to calculate the coordinates of the next grid point. Reference numeral 1307 denotes a temporary storage device that temporarily stores the output of the pixel number conversion device 1303. 1308 is an x coordinate storage device 1305 and y
The address of the memory 1309 that stores the result image is calculated based on the output of the coordinate storage device 1306, and the temporary storage device 1
The pixel transfer device stores the contents of 307 in the memory 1309. Reference numeral 1309 denotes a memory that stores the result image.

The operation of the conventional image rotation device constructed as described above will be described. First, the pixel number conversion device 1303 reads the original image stored in the memory 1301, and reduces the original image in the horizontal direction and in the vertical direction based on the output from the rotation angle holding device 1302, thereby converting the image. The number of pixels is converted, and the result is written in the temporary storage device 1307. The pixel transfer device 1308 inputs one pixel of the image stored in the temporary storage device 1307,
x-coordinate storage device 1305 and y-coordinate storage device 1306
An address of a memory in which the pixel is stored is calculated from the value stored in the memory and stored in the memory 1309. In addition, the grid point coordinate calculation device 1304 is the rotation angle holding device 1.
302, the coordinate of the next grid point is calculated based on the values of the x coordinate storage device 1305 and the y coordinate storage device 1306,
Every time the pixel transfer device 1308 processes one pixel, the contents of the x coordinate storage device 1305 and the y coordinate storage device 1306 are updated. By repeating the above processing, the rotation processing of the two-dimensional rectangular image can be performed.

[0015]

However, the above-described conventional image rotation device has a problem that it is difficult to speed up the rotation process because the rotation image is calculated pixel by pixel. It was Further, since a circuit for enlarging the original image in the vertical direction is required, there is a problem that the circuit amount of the image rotating device becomes large.

The present invention solves the above-mentioned problems of the prior art, and provides an image rotation device capable of performing rotation processing at high speed and not requiring a circuit for vertically expanding an original image. The purpose is to

[0017]

In order to achieve the above object, the present invention provides a first memory for storing an original image and a second memory for storing an image obtained by rotating an original image. , An input processing unit for reading the image data of the original image from the first memory, a horizontal enlargement / reduction processing unit for converting the number of pixels in the horizontal direction of the image data input from the input processing unit, and a rotation process for the original image. An output unit that performs coordinate calculation of transfer grid points and writes image data input from the horizontal enlargement / reduction processing unit to a specified address of the second memory, and an overall control unit that controls each unit of the device Is.

[0018]

According to the present invention, with the above configuration, the overall control unit controls the number of operations of the input processing unit, the lateral enlargement / reduction processing unit, and the output processing unit, thereby eliminating the vertical enlargement circuit of the original image. Image rotation processing can be realized at high speed.

Further, according to the present invention, when the output processing unit arranges each pixel of the result image obtained by rotating the original image at the grid point on the two-dimensional coordinate, the result existing on the same horizontal grid line In order to calculate the number of pixels of the image and write the data stored in the same word of the second memory in word units at a time,
The rotation process can be performed at high speed when the rotation angle θ is −1 ≦ tan θ ≦ 1.

Further, according to the present invention, the overall control unit determines the number of times the horizontal enlargement / reduction processing unit performs the enlargement processing so that one pixel constitutes one word, and the output processing unit writes to the second memory. To realize horizontal reduction processing of the original image, thereby realizing image rotation processing without providing a circuit for performing image data shift processing for writing image data in the second memory. You can

Further, according to the present invention, when the original image is rotated to obtain the result image, among the pixels forming the result image,
By inputting the coordinates of two or more pixels as parameters, the image rotation processing can be performed without performing the trigonometric function operation. Therefore, a high-speed rotation processing can be performed without providing a circuit for performing the trigonometric function operation. Can be done. In addition, the affine transformation can be easily performed by arbitrarily setting the coordinate values of two or more pixels.

[0022]

【Example】

 (Embodiment 1) Hereinafter, a first embodiment of the present invention will be described.
To do. In the first embodiment of the present invention, the rotation angle θ is −1.
The rotation process when ≦ tan θ ≦ 1 will be described. Theory
For the sake of simplicity, we first use FIG.
Rotation processing of one-dimensional image when tan θ ≦ 1
explain. In Fig. 3, each pixel of the original image
The resulting image obtained by being rotated is represented by a black circle, and W is the original image.
Represents the number of pixels. Point A ₀Is the leftmost pixel in the original image, point A
_W-1Represents the rightmost pixel in the original image, and the point A₀Is Hara
In point. In addition, the result image has the origin of rotation (point A
₀), The image obtained when the rotation angle is θ
It Further, when M is defined as follows, M = [Wcos θ] (6) Each pixel of the result image is on each vertical grid line up to x = M−1.
There is one in. Therefore, the number of pixels of the resulting image is M
It becomes an individual. Point B₀, Point B_M-1Are the points of the original image
A₀, Point A_W-1Is a pixel of the result image corresponding to. this
At this time, the rotation angle θ is represented by tan θ = n / m (7) (where, m ≧ n: m, n is an integer of 0 or more)
It is assumed that the straight line L is represented by L: y = (n / m) x (8).

According to the present invention, the one-dimensional image rotation processing in the case of −1 ≦ tan θ ≦ 1 is performed by the pixel number conversion processing for reducing the original image from the original image according to the rotation angle θ and along the straight line of the inclination θ. Pixel transfer processing for calculating the coordinates of the transfer grid points and transferring the reduced images in order to the transfer grid points. In this embodiment, the coordinates of the transfer grid points are calculated by calculating the number of transfer grid points existing on the same horizontal grid line. As a result, pixels to be transferred on the same horizontal grid line can be collectively processed.

The process of calculating the number of transfer grid points existing on the same horizontal grid line will be described below with reference to FIG. FIG. 4 is an enlarged view of the vicinity of the intersection of the horizontal lattice line represented by y = j and the straight line L in FIG. 3, and is the leftmost one of the transfer lattice points on the j-th horizontal lattice line. When the x coordinate of the transfer grid point B _i is XB _i and the y coordinate is YB _i , when the reduced original image is transferred to the transfer grid point calculated from the straight line L, a straight line is formed on the actual digital image. The pixels are arranged so that the y coordinate YB _i of the transfer grid point B _i on the i-th vertical grid line is YB _i = [(n / m) × i] (9). At this time, if α and β such that α = [m / n] (10) β = m−n × [m / n] (11) are defined, transfer lattices existing on the same horizontal lattice line are defined. The number of points can be no more than α or α + 1. At this time, whether the number P _j of transfer lattice points existing on the j-th horizontal lattice line is α or α + 1 is
From the value R _j obtained by integrating the error β due to quantization,
It can be calculated as follows. if R _j + (β-n) <0 P _j = α R _{j + 1} = R _j + β else P _j = α + 1 R _{j + 1} = R _j + (β-n) (12)

When a straight line is drawn based on the equation (9), there are α pixels or α pixels appearing on the same horizontal grid line.
The fact that the number is +1 is disclosed in P3 to 4 of Japanese Patent Application Laid-Open No. 63-142479 filed by the applicant of the present application.

In this way, they exist on the same horizontal grid line.
Calculate the number of transfer grid points to be present and exist on the same horizontal grid line
Coordinates of the leftmost transfer grid point among the transfer grid points
Others that are on the same horizontal grid line can be calculated by
It is no longer necessary to calculate the coordinates of the transfer grid points of
Pixels transferred on the flat grid can be processed collectively.
Wear. For example, the pixel B shown in FIG._iCalculate the coordinates of
For example, pixel B_{i + 1}And B_{i + 2}For, calculate its coordinates
Since it does not need to be output, pixel B_i, B_{i + 1}, B _{i + 2}Of 3
One pixel can be processed at the same time.

Next, the processing procedure for rotating the one-dimensional image in this embodiment will be described step by step. Step A1 A process of reducing the original image composed of W pixels to an image composed of [W × cos θ] pixels is performed. This is shown in Figure 3.
As shown in, the distance between the point B ₀ and the point B _M-1 of the result image obtained by performing the rotation processing is almost the same as the distance between the point A ₀ and the point A _W-1 of the original image. This is to make them equal.

Step A2: The x coordinate of the leftmost transfer lattice point on the jth horizontal lattice line
Standard XF_j, And the y-coordinate YFj, the j-1th horizontal case
The x coordinate XF of the leftmost transfer grid point on the line _j-1, Y
Coordinate YF_j-1, And on the j-1th horizontal grid line
Number of transfer grid points P_j-1From, calculate as follows
It XF_j= XF_j-1+ P_j-1 ... (13) YF_j= YF_j-1+1 (14) However, XF₀= YF₀= 0

Step A3: The number P of transfer grid points existing on the j-th horizontal grid line
Whether _j is α or α + 1 is determined by equation (12).
Calculate from

Step A4 Exists on the j-th horizontal grid line calculated in Step A3
If the number of transfer grid points Pj to be processed is α, then step
The j-th horizontal in the image obtained by reduction in the processing of A1
A process for collectively transferring α pixels to be transferred on the grid line
Do the reason. P _jIs α + 1, step A
The j-th horizontal case in the image obtained by reduction in the process 1
Transfers α + 1 pixels to be transferred onto the child line at once
Perform processing.

Step A5 The processing from step A2 to step A4 is performed for each j satisfying the following equation. This means that when transferring a one-dimensional image having the number of pixels of W to a transfer grid point calculated from the straight line L, the number of horizontal grid lines at which the transfer grid point exists is [W × sin θ] +1. It depends on the reason. 0 ≦ j <[W × sin θ] +1 (15) By performing the above processing, the one-dimensional image rotation processing can be performed.

Next, the rotation processing of the two-dimensional rectangular image will be described with reference to FIG. FIG. 5 is for explaining the rotation process when the two-dimensional image is rotated by the angle θ with the center of rotation as the origin. In FIG. 5, y
A straight line having an intercept k (k is an integer) and an inclination (n / m) is L _k
Express. That is, the straight line L _k is represented as follows. L _k : y = (n / m) x−k (16) Further, the inclination is (−tan (π / 2−θ)) passing through the origin.
Is represented by G. That is, the straight line G is expressed as follows. G: y =-(m / n) x (17)

In the rotation processing of the two-dimensional rectangular image, the transfer start point is calculated as the transfer start point S _k, which is the grid point existing on the vertical grid line closest to the intersection of the straight line G and the straight line L _k .
This can be realized by performing the above-described processing of steps A1 to A5 for each horizontal line of the original image.
The black circles in FIG. 5 represent the transfer grid points calculated from the straight line L ₀ and the transfer start points S _k on the straight line L _k .

In the above step A1, one horizontal line of the original image consisting of W pixels is reduced to an image consisting of [W × cos θ] pixels. For this reason, in order to make the number of pixels of the original image and the rotated result image substantially equal, the processing from the above steps A1 to A5 is performed by [H
/ Cos θ] times. Since the original image is composed of H horizontal lines, in this embodiment, ([H
/ Cos θ] −H) line to the horizontal line of the original image,
By performing the processing from steps A1 to A5 twice, the rotation processing that makes the number of pixels of the original image and the rotated result image substantially equal is realized. The processing will be specifically described below.

Step B1 The x coordinate X of the transfer start point S _k of one horizontal line of the original image
S _k and Y coordinate YS _k are calculated from the following formulas. However,
In this case, one horizontal line of the original image is intended to be transferred to the transfer grid points calculated from the straight line L _k, forward transfer lattice points existing on a nearest vertical grid line from the intersection of the straight line G and the straight line L _k Let S _k be the starting point. XS _k = [kcos θ sin θ + 0.5] (18) YS _k = − [kcos ² θ + 0.5] (19)

Step B2 For one horizontal line of the original image shown in Step B1,
The processing from steps A1 to A5 is performed. However, the transfer start point is S _k .

Step B3 For one horizontal line of the original image shown in Step B1,
It is determined whether or not the processes of steps A1 to A5 are performed twice.

Step B4 When it is judged in the process of step B3 that the process is to be performed only once, the process returns to step B1 and the line next to the horizontal one line of the original image which has been processed up to now is a straight line L _{k +. 1}
The process for transferring to the transfer grid point calculated from is started.

Step B5 When it is judged in the processing of step B3 that the processing is performed twice, one horizontal line of the original image shown in step B1 is
In order to transfer to the transfer grid point calculated from the straight line L _{k + 1,} the transfer start point S _{k + 1} of _one horizontal line of the original image shown in step B1 is calculated from the following formula. XS _{k + 1} = [(k + 1) cos θ sin θ + 0.5] (20) YS _{K + 1} = − [(k + 1) cos ² θ + 0.5] (21)

Step B6 For one horizontal line of the original image shown in Step B1,
The processing from steps A1 to A5 is performed. However, the transfer start point is S _{K + 1} .

Step B7 Returning to step B1, the process for transferring the line next to the horizontal one line of the original image processed up to this point to the grid point calculated from the straight line L _{k + 2} is started.

Step B8 Steps B1 to B7 are performed H times.

As described above, after the transfer start point of each horizontal 1 line of the original image is calculated from the intersection of the straight line L _k and the straight line G, each pixel of the reduced image of the original image is sequentially arranged from the straight line L _k. If it is transferred to the calculated transfer grid point, the rotation processing of the two-dimensional image can be realized. Further, the processing from steps B5 to B7 described above is performed by ([H / cos θ] −
By performing H) times, it is possible to realize the rotation processing that makes the number of pixels of the original image and the rotated result image substantially equal to each other, without requiring a vertical enlargement circuit of the original image.

Next, the structure of the image rotation device for carrying out the above algorithm will be described with reference to FIG.
In FIG. 1, 101 is a first memory that stores an original image. 102 calculates the address of the original image read from the first memory 101 and stores the original image in the first memory 1
This is an input processing unit for inputting from 01. A horizontal enlargement / reduction processing unit 103 enlarges / reduces the original image input from the input processing unit 102 in the horizontal direction. Reference numeral 104 calculates the coordinates of the transfer grid points, and the horizontal enlargement / reduction processing unit 103.
The output processing unit writes the image input from the second memory 105 into the second memory 105. Reference numeral 105 denotes a second memory that stores the rotated result image. Reference numeral 106 denotes the input processing unit 10.
2, an overall control unit that controls the horizontal reduction processing unit 103 and the output processing unit 104. 107, 108, and 109 are output from the overall control unit 106, and are input control unit 10 respectively.
2, a control signal for controlling the horizontal enlargement / reduction processing unit 103 and the output processing unit 104. Reference numerals 110 to 113 are data lines, and each processing unit and memory inputs / outputs image data via the data lines.

Next, the output processing unit 104 in FIG. 1 will be described in detail with reference to FIG. In FIG. 2, 201
Is an image temporary storage device for temporarily storing the image from the horizontal enlargement / reduction processing unit 103. Reference numeral 202 denotes an output operation device that performs an operation on the data read from the second memory 105 via the data line 113 and writes the result image output from the image temporary storage device 201 to the second memory 105. Reference numeral 203 denotes an address generation device that generates an address for writing the resulting image data in the second memory 105. An output processing unit control device 204 controls the output processing unit 104. 205, 206, 207
Is a control signal output from the output processing unit control device 204 and controls the image temporary storage device 201, the output operation device 202, and the address generation device 203, and 208 is a data bus.

Next, the operation of the image rotation device configured as described above will be described. Introduction Overall control device 10
In step 6, the process of step B1 is performed, and the input processing unit 10
2, control signals 107, 1 so that the horizontal enlargement / reduction processing unit 103 and the output processing unit 104 can perform the processing of step B2.
Each processing unit is set via 08 and 109.

The input processing unit 102 generates an address for reading the image data corresponding to one horizontal line of the original image from the first memory 101, and the image data of one horizontal line is generated in the first memory according to the address. The image data is read from 101 and the read image data is output word by word to the horizontal enlargement / reduction processing unit 103. Horizontal enlargement / reduction processing unit 10
In step 3, the processing shown in step A1 is performed on the pixels included in the image data of 1 word input from the input processing unit 102, and the reduced image is output word by word in the image temporary storage device of the output processing unit 104. Output to 201. In the output processing unit 104, the address generation device 203 performs steps A2, A3, and A4 to calculate the address of the second memory 105 in which the image data output from the horizontal enlargement / reduction processing unit 103 is stored. Address second
Output to the memory 105 of
2 to output the image data stored in the image temporary storage device 201 to the second memory 105. At this time, the second
When it is necessary to perform calculation with the data stored in the memory 105, the data is once read from the second memory 105, and the output calculation device 202 uses the image temporary storage device 20.
After performing an operation with the data stored in 1, the data is stored in the second memory 105 again.

In the overall control unit 106, the input processing unit 10
2. While the processing units of the horizontal enlargement / reduction processing unit 103 and the output processing unit 104 are performing the above-described processing for one horizontal line of the original image, the processing of step B3 is performed. If the horizontal 1 line of the original image is a line that is processed only once, the process of step B1 is performed for the process of the next line. 1 horizontal line of the above original image is 2
In the case of a line to be processed once, the process of step B5 is performed, and the set value set in each processing unit is calculated. When the processing for one horizontal line of the original image is completed, the set values calculated in the respective processing units of the input processing unit 102, the lateral enlargement / reduction processing unit 103, and the output processing unit 104 are set as the control signals 107 and 108. , 109 to start the next process. The above process is performed based on step B8.
Repeat.

As described above, according to the first embodiment, if only the coordinates of the leftmost transfer grid point existing on the horizontal grid line are calculated, the other coordinates existing on the same horizontal grid line are calculated. Since it is not necessary to calculate the coordinates of the transfer grid points, a plurality of pixels can be collectively written from the output processing unit 104 to the second memory 105, and the number of memory accesses that occur during rotation processing can be reduced. it can. For example, in FIG.
As shown in, when one word is composed of 4 pixels, (a1, a2, a3, a4), (a5, a6, a
Image data of 2 words composed of pixels 7 and a8) is output in units of words from the horizontal enlargement / reduction processing unit 103 to the output processing unit 104, and each pixel is output to the second memory 10.
Stored in 5. On the other hand, in the conventional image rotating apparatus, since the pixels are written in the memory one pixel at a time, for example, in order to write the pixels a1, a2, and a3 in the memory, three memory accesses are required. On the other hand, in the present embodiment, since this can be realized by one memory access, the image rotation processing can be speeded up.

In addition, according to the present embodiment, the pipeline configuration is adopted, and a plurality of pixels included in one word can be collectively processed in one machine cycle in each stage. Further, the image data of 1 word can be output to the memory, and the image rotation processing can be speeded up.

Furthermore, according to the present embodiment, since the rotation processing can be performed without providing a circuit for enlarging the original image in the vertical direction, the circuit amount of the image rotating device can be reduced.

In this embodiment, the transfer grid point is given by the equation (9), but the process of setting the grid point closest to the intersection of the straight line L and the vertical grid line as the transfer grid point can be easily realized. You can Further, in the present embodiment, the center of rotation is the upper left point of the original image, but it can be easily inferred that an arbitrary position can be the center of rotation.

(Embodiment 2) Next, a second embodiment of the present invention will be described. In the second embodiment of the present invention, a rotation process when the rotation angle θ is tan θ <−1 or tan θ> 1 will be described. The rotation angle θ is tan θ <−
1 or tan θ> 1, the reduced original image is converted into a straight line L
, The x-coordinate XB _j and the y-coordinate Y of the pixel B _j on the j-th horizontal grid line on the actual digital image.
The pixels are arranged such that B _j is XB _j = [(n / m) × j] (22) YB _j = j. An example of this is shown in FIG. FIG. 7 shows the arrangement of the pixels B _j in 0 ≦ j ≦ 7.

The rotation angle θ is tan θ <−1 or ta
When nθ> 1, only one transfer grid point appears on the same horizontal grid line. Therefore, even if the processing of steps A1 to A5 shown in the first embodiment is performed to calculate the coordinates of the transfer grid points existing on the same horizontal grid line,
Since the coordinates of each transfer grid point must be calculated, the image rotation processing cannot be speeded up.
Therefore, in this case, the coordinates of the transfer grid point are calculated from the equation (22) and the pixels are transferred pixel by pixel, without performing the processing from steps A1 to A5 shown in steps B2 and B6. Note that step B1, steps B3 to B5,
Regarding steps B7 and B8, the operation is the same as in the first embodiment.

The configuration of the image rotation device in the second embodiment of the present invention is the same as that of the first embodiment shown in FIGS. 1 and 2, but the operation is different from that of the first embodiment. The operation of the second embodiment will be described below. First, the overall control device 106 performs the process of step B1, and the input processing unit 102, the lateral enlargement / reduction processing unit 103, and the output processing unit 104.
So as to calculate the coordinates of the transfer grid point from the equation (22) and transfer the pixel by pixel, the control signal 107,
Each processing unit is set via 108 and 109.

The input processing unit 102 generates an address for reading the image data corresponding to one horizontal line of the original image from the first memory 101, and outputs the image data of one horizontal line from the image memory 101 according to the address. The horizontal enlargement / reduction processing unit 103 that reads the read data
1 pixel at a time.

In the horizontal enlargement / reduction processing unit 103, all the pixels of one word output to the output processing unit 104 are enlarged to the pixels input from the input processing unit 102.
The result is stored in the image temporary storage device 201 of the output processing unit 104.
Output to. For example, when one word is composed of 4 pixels and the pixel input from the input processing unit 102 is represented as a1, the horizontal scaling processing unit 103 performs 4 times expansion and is included in one word. Enlargement processing is performed so that all pixels become a1.

In the output processing unit 104, the image data input from the horizontal enlargement / reduction processing unit 103 is stored in the image temporary storage device 2.
Store in 01. Then, based on the control signal 109 input from the overall control unit 106 to the output processing unit control device 204, it is determined whether to output the data stored in the image temporary storage device 201 to the second memory 105.

When outputting, the address generator 20
Equation (22) is calculated in step 3, and the image temporary storage device 20
1 calculates the address for storing the image data stored in No. 1, outputs the address to the second memory 105, and outputs the image from the output arithmetic unit 202 to the image temporary storage unit 20.
The image data stored in 1 is output to the second memory 105, and one pixel that has been the target of processing is written in the memory. When not outputting, the operation of storing in the second memory 105 is not performed, and the process for the next image data input from the lateral enlargement / reduction processing unit 103 is started.

In this embodiment, unlike the case of the first embodiment, in the overall control unit 106, each processing unit of the input processing unit 102, the lateral enlargement / reduction processing unit 103, and the output processing unit 104 has the above-mentioned original image. While the processing for 1 horizontal line is being performed, the calculation for controlling the number of data output from the image temporary storage device 201 to the second memory 105 is performed as described above. As a result, the step A performed by the horizontal enlargement / reduction processing unit 103 in the first embodiment is performed.
1 horizontal reduction processing is realized.

When the process for one horizontal line of the original image is completed, the process of step B3 is performed. If the horizontal 1 line of the original image is a line that is processed only once, step B1 is performed for processing the next line.
Is processed. When the horizontal 1 line of the original image is a line to be processed twice, the process of step B5 is performed, the parameters set in each processing unit are calculated, and the input processing unit 102, the horizontal enlargement / reduction processing unit 103, Output processing unit 104
The respective processing units are set, and the next processing is started. The above process is repeated based on step B8.

As described above, this embodiment is different from the operation of the first embodiment in two points. One is that the horizontal enlargement / reduction processing unit 103 performs enlargement processing of one pixel, and outputs a one-word image formed by all the pixels of one word to the image temporary storage device 201. The second is that the horizontal reduction processing of the original image is realized by controlling the number of data output by the image temporary storage device 201 from the overall control unit 104.

The operation in the second embodiment will be described below as the first operation.
The reason why the operation is different from that of the embodiment will be described. When the rotation angle θ is tan θ <−1 or tan θ> 1, there are two or more transfer grid points on the same vertical grid line as shown by points B ₀ and B ₁ in FIG. In order to transfer the pixels of the original image to this transfer grid point, it becomes necessary to align the pixels. An example is shown in FIG. Figure 8
Is represented by the values of pixels at points B ₀ and B ₁ in FIG. 7 as a ₀ and a ₁ , respectively, and each pixel is stored in the first pixel and the next pixel of one word in the memory storing the original image. As a result of the rotation processing, each pixel is stored as the first pixel of one word. As shown in FIG. 8, when processing the point B ₁ , it is necessary to change the position of the pixel of the original image. In this embodiment, the alignment of the pixels of the original image is realized without providing a shift circuit. That is, the horizontal enlargement / reduction processing unit 103 performs enlargement, and all the pixels of one word are set as a 1 which is the value of the pixel of the point B ₁ and the image temporary storage device 201 of the output control unit 104.
By performing the output to, the pixel alignment is realized without providing a shift circuit.

As described above, according to the present embodiment, it is not necessary to provide a shift circuit for aligning the pixels when writing the processed pixels in the memory, so that the circuit amount of the image rotation device is reduced. be able to.

(Embodiment 3) Next, a third embodiment of the present invention will be described. In the present embodiment, an image rotation device that can perform image rotation processing without performing trigonometric function calculation will be described. The configuration of the image rotation device of this embodiment is the same as that of the first and second embodiments.
The operation of the present embodiment is different from the first and second embodiments in the horizontal reduction processing of the original image, the vertical enlargement processing, and the calculation of the transfer start point, but the other operations are performed. This is the same as the first and second embodiments.

Regarding the operation of this embodiment, first, the horizontal reduction processing and the vertical enlargement processing of the original image will be described. In the present embodiment, when performing the image rotation process with the upper left pixel of the original image as the center of rotation, the x coordinate of the upper right pixel of the resulting image is M, the y coordinate is N, and M and N are the main image rotations. The rotation processing is realized by inputting to the processing device. At this time, by giving M and N by the following formulas, it is possible to realize the rotation processing in which the rotation angle is substantially equal to θ. M = [Wcos θ] (23) N = [W sin θ] (24)

As described above, by inputting the above M and N to the image rotation device, it is not necessary to perform the trigonometric function calculation in the following three processes shown in the first and second embodiments. The first is equation (10) and equation (11).
Calculation of α and β. To obtain α and β,
It is necessary to obtain m and n defined by the equation (7).
However, in the present embodiment, α and β are calculated by using M and N given by Equations (23) and (24), and α = [M / N] (25) β = M−N × [ M / N] (26), the image rotation processing is realized. At this time, in order to perform the two-dimensional rotation processing, the straight line L _k and the straight line G shown in FIG. 5 are given by the following equations. L _k : y = (N / M) x−k (27) G: y = − (M / N) x (28)

The second is the horizontal reduction processing in step A1 of the first embodiment. In step A1, an original image composed of W pixels is reduced to an image composed of [W × cos θ] pixels. In the present embodiment, the processing of reducing the original image composed of W pixels into an image composed of M pixels by the equation (23) is sufficient.
It is not necessary to calculate the trigonometric function in the process of step A1.

The third is the process in step A5 of the first embodiment. In step A5, the processing from step A2 to step A4 is performed [W × sin θ] +1 times. In the present embodiment, since the processing from step A2 to step A4 may be performed N + 1 times by the equation (24), the trigonometric function may not be calculated in the processing of step A5.

Next, the calculation of the transfer start point is carried out.
The operation in the example will be described. The first embodiment and the second
In the embodiment, the total of formulas (18) and (19) in step B1 is calculated.
Calculate the straight line L_kIs calculating the transfer start point in
However, in this embodiment, the straight line L _kTransfer start point S in_k
X coordinate XS_k, Y-coordinate YS_kWhen is calculated,
Straight line L_{k + 1}Transfer start point S in_{k + 1}X coordinate of
S_{k + 1}, Y-coordinate YS_{k + 1}Is calculated as follows. However
And XS₀= YS₀= 0.

If rs _k + N / M ≦ 1 (29.1) rs _{k + 1} = rs _k + N / M (29.2) rm _{k + 1} = rm _k (29. 3) XS _{k + 1} = XS _k ... (29.4) YS _{k + 1} = YS _k -1 (29.5) else rs _{k + 1} = rs _k + N / M-1 ... (29.6) if rm _k + N / M ≦ 1 (29.7) rm _{k + 1} = rm _k + N / M (29.8) XS _{k + 1} = XS _k +1 (29.9) YS _{k + 1} = YS _k -1 (29.10) else rm _{k + 1} = rm _k + N / M-1 (29.11) XS _{k + 1} = XS _k +1 ... (29.12) YS _{k + 1} = YS _k ... (29.13)

The above processing will be described in detail with reference to FIG.
It In the above process, rs_kIs the transfer start point S_kof
Fractional part of the x coordinate of the intersection of the existing horizontal grid line and straight line G
Represents minutes. XS₀= 0, the straight line G passes through the origin.
So rs₀= 0. Also rm_kIs the transfer starting point
S_kVertical grid line and straight line L where_kY coordinate of the intersection point with
Represents the fractional part of. YS₀= 0 and the straight line L₀
Passes through the origin, so rm ₀= 0.

Such rs_kAnd rm_kIs given
First, when using the equation (29.1), rs_kOf straight line G
By adding the reciprocal of the slope, the straight line G becomes y = YS
_kAnd y = YS_kWhether to cross the vertical grid line between -1 and
Determine whether. When the straight line G does not cross the vertical grid line
To change the x-coordinate of the transfer start point in equation (29.4).
First, the y coordinate is set to −1 by the equation (29.5), and rs is set._kEquation (2
9.2), rm_kLet be the value given by (29.3)
It If the straight line G crosses the vertical grid line, rs _kThe formula
The value is given by (29.6). Next, equation (29.
In 7), rm_kStraight line L_kBy adding the slope of
Straight line L_kIs x = XS_kAnd x = XS_kHorizontal between +1
Judge whether to cross the grid line. Straight line L_kIs horizontal
If it does not exceed the grid line, transfer starts with equation (29.9).
The x coordinate of the point is incremented by +1 and the y coordinate is decremented by -1 by the equation (29.10).
And then rm_kBe the value given by equation (29.8). straight
Line L_kIs above the horizontal grid line, the equation (29.1)
In step 2), the x-coordinate of the transfer start point is incremented by 1 to obtain the equation (29.13).
Does not change the y coordinate, and rm_kIs given by formula (29.11)
It should be a value that can be obtained.

By performing the above processing in this way, the triangle
Straight line L without using function calculation_kTransfer start point S in
_kCan be calculated. In FIG. 9, a straight line L_kof
Transfer start point S_kIs the straight line L_k-1Transfer starting point S_k-1Zodiac
From equation (29.4) and equation (29.5),
It shows an example that can be obtained by. Also, the straight line L
_{k + 1}Transfer starting point S_{k + 1}Is the straight line L_kTransfer starting point S_k
From the coordinates of, equation (29.9) and equation (29.10)
It shows an example that can be obtained by going. further,
Straight line L_{k + 2}Transfer starting point S_{k + 2}Is the straight line L_{k + 1}Transfer opening
Starting point S_{k + 1}From the coordinates of, the equation (29.12) and the equation (2
It shows an example that can be obtained by performing
It

As described above, according to the present embodiment, the image rotation processing can be performed without performing the trigonometric function operation, so that an arithmetic device capable of performing the trigonometric function operation is not required and the image rotation is performed. The circuit scale of the device can be reduced, and the calculation time required for the calculation of trigonometric functions can be shortened.

The equation (29.
The series of processing from 1) to equation (29.13) is [H / c
[os.]] times, the formula (29.11) to the formula (29.
The processes up to 13) occur ([H / cos θ] −H) times. This is equal to the number of horizontal lines of the original image in which the processing of steps A1 to A5 shown in the first embodiment is performed twice just to enlarge the original image in the vertical direction.
That is, from the above equation (29.1) to equation (29.13)
[H / cos θ] times, and the formula (2
Regarding the processing from 9.11) to the equation (29.13), if it is performed for one horizontal line of the same original image as the immediately preceding processing, the calculation for vertically enlarging the original image is performed. There is no need.

Further, in the present embodiment, when performing the image rotation process with the upper left pixel of the original image as the center of rotation, the x coordinate of the pixel of the result image corresponding to the upper right pixel of the original image is M, y. Although the coordinates are N, this is as shown in FIG. 10, where the x coordinate of the lower left pixel of the original image is M ′, the y coordinate is N ′, and M ′ and N ′ are input to the main image rotation processing device. Therefore, the same effect can be obtained. At this time, by giving M ′ and N ′ by the following expressions, it is possible to realize the rotation processing in which the rotation angle is substantially equal to θ. M ′ = [H sin θ] (30) N ′ = [H cos θ] (31) Even if all of M, N, M ′, and N ′ are input to the image rotation processing device, The same effect can be obtained. Further, even in an image rotation device different from the configuration of the image rotation processing device of the present invention, the calculation for the rotation process is simplified by inputting the above M, N, M ′ and N ′ to the image rotation processing device. What you can do is easy to guess.

Further, in the present embodiment, the above M, N,
M ′ and N ′ are represented by equations (23), (24), and (3
0) and (31), but M, N, M ′,
By setting N ′ arbitrarily, affine transformation of a rectangular image can be easily realized. In the affine transformation, there is a case where the original image needs to be enlarged in the horizontal direction and reduced in the vertical direction, but in the present embodiment, the reduction in the vertical direction controls the memory access of the input processing unit 102, This can be realized by reducing the number of horizontal lines of the original image read from the first memory 101. Further, horizontal enlargement can also be realized by the horizontal enlargement / reduction processing unit 103 in the case of the first embodiment, and even in the second embodiment, the same data is stored in the memory multiple times from the output processing unit 104. Since it can be realized by writing, it can be easily inferred from this that the present embodiment can realize the affine transformation.

Although the center of rotation is the origin in each of the first, second and third embodiments of the present invention, if the coordinates of the center of rotation are input to the image rotation device. ,
It can be easily inferred that the center of rotation of the image can be set at an arbitrary position.

[0080]

As described above, according to the present invention, the first memory for storing the original image, the second memory for storing the image obtained by performing the rotation processing of the original image, and the first memory An input processing unit that reads image data from, a horizontal enlargement / reduction processing unit that inputs image data from the input processing unit and converts the number of pixels in the horizontal direction of the original image, and performs coordinate calculation of transfer grid points and horizontal enlargement / reduction. An input processing unit, a lateral enlargement / reduction processing unit, and an output processing unit, and an output processing unit that writes image data input from the unit to a specified address of the second memory and an overall control unit that controls each unit of the device. By controlling the number of times of operations of the above by the overall control unit, the image rotation processing can be realized at high speed without providing a vertical enlargement circuit for the original image.

Further, according to the present invention, when the output processing arranges each pixel of the result image obtained by rotating the original image at the grid point on the two-dimensional coordinates, the result image existing on the same horizontal grid line Is calculated and the data stored in the same word of the second memory is written at a time on a word-by-word basis, it is possible to perform a rotation process at a rotation angle θ of −1 <tan θ ≦ 1 at high speed. it can.

Further, according to the present invention, the overall control unit controls the number of times the enlargement processing is performed so that the pixels processed by the horizontal enlargement / reduction processing unit form one word, and the writing is performed from the output processing unit to the second memory. To realize horizontal reduction processing of the original image, thereby realizing image rotation processing without providing a circuit for performing image data shift processing for writing image data in the second memory. You can

Further, according to the present invention, when the original image is rotated to obtain the result image, among the pixels forming the result image,
By inputting the coordinates of two or more pixels as parameters, the image rotation processing can be performed without performing the trigonometric function operation. Therefore, a high-speed rotation processing can be performed without providing a circuit for performing the trigonometric function operation. Can be done. In addition, the affine transformation can be easily performed by arbitrarily setting the coordinate values of two or more pixels.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of an image rotation device according to a first embodiment, a second embodiment, and a third embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration of an output processing unit in the device.

FIG. 3 is a conceptual diagram for explaining a rotation process of a one-dimensional image according to the first embodiment of the present invention.

FIG. 4 is a conceptual diagram for explaining a rotation process of a one-dimensional image according to the first embodiment of the present invention.

FIG. 5 is a conceptual diagram for explaining a two-dimensional image rotation process in the first embodiment of the present invention.

FIG. 6 is a conceptual diagram for explaining an effect in the first embodiment of the present invention.

FIG. 7 is a conceptual diagram for explaining a rotation process of a one-dimensional image according to the second embodiment of the present invention.

FIG. 8 is a conceptual diagram for explaining the effect of the second embodiment of the present invention.

FIG. 9 is a conceptual diagram for explaining a method of calculating a transfer start point according to the third embodiment of the present invention.

FIG. 10 is a conceptual diagram for explaining an input value to the image rotation device in the third embodiment of the present invention.

FIG. 11 is a conceptual diagram for explaining a one-dimensional image rotation process of a conventional image rotation device.

FIG. 12 is a conceptual diagram for explaining a two-dimensional image rotation process of a conventional image rotation device.

FIG. 13 is a block diagram showing a configuration of a conventional image rotation device.

[Explanation of symbols]

 101 First Memory 102 Input Processing Unit 103 Horizontal Enlargement / Reduction Processing Unit 104 Output Processing Unit 105 Second Memory 106 Overall Control Unit 107 Control Signal 108 Control Signal 109 Control Signal 110 Data Line 111 Data Line 112 Data Line 113 Data Line 201 Image temporary storage device 202 Output operation device 203 Address generation device 204 Output processing unit control device 205 Control signal 206 Control signal 207 Control signal 208 Data bus

Continuation of the front page (72) Inventor Shigeo Shimazaki, 1006 Kadoma, Kadoma City, Osaka Prefecture Matsushita Electric Industrial Co., Ltd.

Claims (6)

[Claims] 1. A first memory for storing an original image, a second memory for storing an image obtained by rotating an original image, and an input for reading image data of the original image from the first memory. A processing unit, a horizontal enlargement / reduction processing unit that converts the number of pixels in the horizontal direction of the image data input from the input processing unit, coordinate calculation of transfer grid points for rotation processing of the original image, and horizontal enlargement processing. An image rotation device comprising: an output processing unit for writing image data input from a reduction processing unit to a specified address of the second memory; and an overall control unit for controlling each unit of the device. 2. An image temporary storage device in which the output processing unit temporarily stores the image data input from the horizontal enlargement / reduction processing unit,
An output computing device that outputs the image data input from the image temporary storage device to a second memory, and an address that calculates and outputs an address for storing the image data output by the output computing device in the second memory The apparatus according to claim 1, further comprising: a generation device;
The image rotation device described. 3. The output processing unit has a rotation angle θ of −1 ≦ ta.
The number of pixels of the result image existing on the same horizontal grid line when each pixel of the result image obtained by performing the rotation process of the original image given by nθ ≦ 1 is arranged at the grid point on the two-dimensional coordinates. 3. The image rotation device according to claim 1, wherein the data stored in the same word of the second memory is written in word units. 4. The rotation angle θ is tan θ <−1 or ta.
When performing the rotation processing of the original image given by nθ> 1,
The horizontal enlargement / reduction processing unit enlarges one pixel of the image data input from the input processing unit to one word, and the output processing unit stores the image data input from the horizontal enlargement / reduction processing unit in the second memory. 3. The horizontal reduction processing of the original image is performed by determining whether or not to write in the image and controlling the number of writing processings.
The image rotation device described. 5. When the original image, which is a rectangular image, is rotated to obtain the result image, instead of using the rotation angle θ, among the four pixels of the result image for the four pixels that represent each vertex of the original image, The image rotation device according to claim 1, wherein coordinate values of two or more pixels are used. 6. The image rotation apparatus according to claim 5, wherein the affine transformation is performed by arbitrarily setting the coordinate values of the four pixels of the resultant image with respect to the four pixels that represent each vertex of the original image. .