JPS6016767A - System for processing rotation of picture - Google Patents

Abstract

PURPOSE:To attain the rotation processing of a picture at a high speed and also to reduce the memory capacity by storing the data of an input picture while converting it into a lateral and a longitudinal compressing codes and restoring the rotation data of the input picture from each compressing code. CONSTITUTION:The picture data inputted to an input terminal 31 is inputted to a converting circuit 34 while being delayed for one line's share by a shift register 32 and converted into the lateral compressing code. Further, the picture data subjected to a delay for one line's share by a shift register 33 is supplied to a deciding circuit 37 and the longitudinal compressing code is formed by a converting circuit 40 through the operation of a memory 39 and an adder 38. These compressing codes are stored in memories 35,41 and the picture data as a result of rotation of the input picture by 0 deg.,90 deg.,180 deg.,270 deg. is restored from these compressing codes by a restoring circuit 43.

Description

[Detailed description of the invention] Industrial applications The present invention relates to a method for image rotation processing.

To the configuration of the conventional example and its problems One of the processes is the rotation process JJ. FIG. 1 (is a small diagram of an example of the rotation process. In this figure, I C:j
, a human image (00 rotation image), 2 is a 90° rotation direction of the human image, 3 is a 1800 rotation image of the input image, and 4 is a 270 degree rotation image of the human image.

FIG. 2 is a schematic block diagram of a small-scale image processing system that performs such image processing such as rotation processing. In this figure, 11idHcPU, 12 is an image display device, 13 is an image input device, 14 is an image memory, and 16 is an image output device.

The binary input image read by the image input device 8'13 is temporarily stored in the image memory 14.

Rotation processing for this input image is executed by software processing in the CPU 11, and the generated rotated image is transferred to the image output device 16, the image display device 12, or the image memory 14.

FIG. 3 is an explanatory diagram of the 900-rotation process, where 21 is an input image (8×8 bits) and 22 is both 900-rotation images. Each hint from 1 to 64 of the input image 21 is converted into a 9Q0 rotated image 22
A 900-revolution image V4: 22';s: is generated by sequentially moving the image to the illustrated device in the space of . However, many such bit operations are performed by software processing. To repeat IJ (64 times in this example),
It takes a considerable amount of time. This example (d, image information amount is 8×
Since it is 8 bits, the processing time +d is relatively short, but it generally requires an enormous amount of processing time. For example, 8doc)
10cm x 10C read with /mm resolution
Since the nr image is 800 x 800 bits of information, the rotation process requires a significantly long time with the processing power of a normal microcomputer, making it unusable and impractical. .

In order to deal with this L processing time problem, the image %
There is a method in which all image rotation processing is performed using processing-dedicated 7-ware. However, such nodeware is large-scale and expensive, making it impossible to install it in a small-scale image processing system due to cost considerations. Moreover, since the anotocol rhythm of the rotation process is similar to that described above, there is a problem that the effect of reducing the processing time is small compared to the coasting process.

In addition, there is also the problem that a large capacity memory capable of storing at least one input image data is required, regardless of whether the software or the 71-ware is used.

Purpose of the Invention The present invention solves the above-mentioned conventional problems.
It is an object of the present invention to provide a new image rotation processing method that can perform image rotation processing at high speed using inexpensive hardware and can reduce the memory capacity for storing input image data. - Structure of the Invention The present invention does not directly perform rotation processing on an input image, but converts input image data into a compression code in the horizontal direction and a pressure code in the inspection vertical direction and stores them.
oo or 180 of the input image from its horizontal compression code
The image rotation process is achieved by restoring the data of the 0 rotation image and also restoring the data of the 90° 17cl-j2700 rotation image of the input image from the vertical compression code.

, Description of Embodiments Embodiments of the present invention will be described below with reference to the drawings. FIG. 4 is a block diagram of an image rotation processing system according to an embodiment of the present invention. In this figure, 31 is an input introduction where image data of a human image is input serially; 32 and 33
are soft registers each for storing one line of image data.

Note that the step-by-step scanning of image data is performed with the horizontal direction as the main scanning direction. 34 is a conversion circuit for converting image data into a horizontally compressed code, and in this embodiment converts it into a well-known MH code string. Reference numeral 35 represents a memory for storing horizontal compression codes, and reference numeral 36 represents an address control circuit for the memory 36.

Each block from 37 to 40 is for compressing image data in the vertical direction. Reference numeral 37 denotes a determination circuit that determines, for each focus, whether the image data output from the soft register 32 and the image data obtained by delaying it by one line by the shift register 33 (image data of the processing line) match, and the determination result. and the value of the processing line image data at that time is output.

38 is an adder, which adds '1''f to the stored value of the corresponding address in the memory 39 for each bit of image data, when the judgment result of the judgment circuit 37 is a match °'; Write to the address.

When the judgment result is ``mismatch'', the value obtained by adding 1'' to the value stored at the corresponding address in the memory 39 (the length of the white or black run in the vertical direction) is output to the conversion circuit 40, and the address Clear the stored value to “°0゛′. The vertical compression code of the image data is output from the conversion circuit 4o. In this example, a well-known MH code string is output as a compressed code.

41 is a memory for storing vertically compressed codes;
2 is an address control circuit for it.

43 is the input image 00.90o, 18002aO from the compression code stored in the memory 35 or 41.
This is a restoration circuit that restores image data of a rotated image of 0. 44 is a node that interfaces with the system CPU (not shown).

Next, the operation will be explained.

The image data inputted to the input terminal 31 is delayed by one line by the shift register 32 and inputted to the conversion circuit 34, where it is converted into a compressed code in the horizontal direction. This compressed code is address: 1ill Address 1l of m circuit 36
By ill mll, Noshiri 3 and others (〃=1
i1njitono chireru.

There was a delay in register 32! 1, the image data is input to f:/7h register 33 (the image data delayed by 1 digit is input to f4J constant circuit 3γ), 1, that is, the line of a certain line and its tide. Corresponding bits of the image data are simultaneously input to the determination circuit 37. Incidentally, each section 37 to 41 does not operate until all the image data of the first line is stored in the shift register 33. All addresses in memory 39 have been cleared to "0°".

1j no 1) J Niuni Rl concubine 2 Rhino's 11th stroke II
When the corresponding bit of the '7 pin data is input to the judgment circuit 37p, the pressure part processing of the force surface starts in 4.1.
1, :V" figure (〆The case where the image data shown in this figure is manually generated, pressure 4.11) ↓■ψ will be explained.

9t is in the process of adding 1 to 1 (bow 1,
The image data of line/1 and line 2 are sequentially input to the determination circuit 37. Since 1 bit 1 is both '°0°' (white), the adder 38 inputs the hit 11 (two corresponding j'' at L/y, )Md l;i?, ill' in the memory 39.
°“1°′ to i υII , :', fr (Ill,
%-Soen's Ato 1/S in armor Lro Hino l-2,
The same applies to 3, s, and 6°7. Bit 4 is jiin 1°
1°' (black), line 2 is "o"', so the addition G'J device 3
8 (the value obtained by adding “°1°” to the ?ttlXfu+ of at1/s corresponding to bit 4 of the ζL memory 39, that is,
Clear the address of 'Complete change JS1 unit 40-\output 'L2, to 0'). , and human 8 as well.

In the process of input 2, the judgment result of d2, bits 1, 5, and 6°T is °゛small match ``'', and the value ``par 2'' for each is sent from the adder 38 to the conversion circuit 40, and is stored in the memory 39. The corresponding address is cleared.

Memory 39 address storage 1 corresponding to bits 2 and 3
The direct value becomes PA2°', and the stored value C', '°1°'C (two 1s, the conversion circuit 40 receives from the adder 38 The given weight (vertical white i ft is the height of the black run) and the processing line image data of 1!h (given by the judgment circuit 37 by breaking the force 1 force 38). (Not compatible (MH 11 to 1, above) and m et al.
ltil circuit 42C'coil address 1ii111i1
Note by!・41 ni Hisoto (i7.'Okibetsu ni l1
ll'+ is stored next.

3j1.6 figure iq+, ・4 le 6 image data & memory in the middle of small direction compression 1 place 3.9.41
This is a schematic diagram of ;e l:1 data. In this figure, d is the point at which the processing of line 2 is completed at r1-, the figure shows 11 points after processing line 3, and line C shows line 4.
The stored data at the time when the processing of is completed is shown.

The data stored in memory 36.41 in this way
11 code, the restoring circuit 43 converts the incoming direction image to all 00.
Image data of 13 images rotated by 90', 180 degrees, or 2700 degrees is restored.

That is, when performing the oO rotation process, the compressed codes of lines 1 to 8 are sequentially read out from the memory 35 under the control of the address 15 and 1 river path 36. Does the restoration circuit 43 decode and decompress those compressed codes? The image data of each line of both images rotated by 0° is restored.

In the case of 1800 rotation processing, lie/8 to 1 from memory 35
The compressed codes are read out in this order.

The decoding circuit 43 decodes those compressed codes (l (
By expanding the image and rearranging it in reverse order, both image data of the 180° rotated image (81 Tanirai/) are restored.

9oo rotation processing, address 1iill cape circuit 4
IN, f of hits 8 to 1 from the memory 41 under the control of
The i14 code is read in this first order! climb. tsj(
-The circuit 43 decodes and expands the compressed code of -Sonera5.
Restoring the image data of each line of the 900 rotation image, 2
When performing a full 700-rpm cleaning, select the pin 1- from memory 41.
The compressed codes i to 8 are sequentially read out in this order.

Restoration circuit 43 (d decodes and decompresses all of those compressed codes,
Furthermore, the image data of each line of the rotated image is restored by arranging and renaming the images in reverse order.

Note that the rearrangement of the bits of the image data in the rotation processing of 1800 and 2700 is not performed by the converter 11 circuit 43, but by C
It may also be performed by software processing in the PU. Such sorting of bits can be executed at a sufficiently high speed by software processing. It is also possible to keep the read order of the compressed codes from the memory 35 and 41 constant and to rearrange the order of the image data of each image restored by the restoration circuit 43 by software processing as necessary. Processing can also be executed at high speed.

Effects of the Invention The present invention converts input image data into horizontal and vertical compression codes and stores them.
Restore the zero-rotation image data from the horizontal compression code, 9
Since this is a method for restoring 0° or 2700 rotation image data from a vertically compressed code, the following effects can be obtained.

(iJ Image rotation processing can be performed at high speed simply by using small-scale and inexpensive hardware such as the one shown in the actual example above.

(10 Since the input image data is compressed and then stored, the memory capacity for this can be significantly reduced. For example, assuming that the horizontal and vertical compression ratios are %, the memory capacity is the same as the input image data. 215 (% ×
2) It can be sharpened.

(Nl) From (1) and (11) above, an image processing system capable of high-speed image rotation processing can be realized at low cost.

[Brief explanation of drawings]

FIG. 1 is a schematic diagram showing an example of image rotation processing; first, a schematic block diagram showing an example of a conventional image processing system; FIG. 3 is an explanatory diagram of 900 rotation processing by a conventional image processing method;
FIG. 4 is a schematic block diagram of a device using an image rotation processing method according to an example of the present invention, FIG. 6 is a schematic diagram showing an example of an input image, and FIG. 6 is a vertical direction in the image rotation processing device described above. FIG. 3 is a schematic diagram showing the transition of data stored in a memory related to compression processing. 23.33...Fft register, 34.40.
... Conversion circuit, 35, 39, 41 ... Memory, 36°42 ... Address control circuit, 37.
...Judgment circuit, 38...Adder, 43...
...Restoration circuit. Name of agent: Patent attorney Toshio Nakao and 1 other person
1 Figure (0゛ @turn) (90°@!f,) (P go'TgJ turn C270' @shaun) Figure 2f Figure 3 9 and turn

Claims (1)

[Claims] Convert the human image data into horizontal compression code and contraction direction compression code and record them as follows: 1-: j+l
: From the horizontal compression code 1 - 0° rotation of the input image
, 1,000 rotations of both images' data gold is restored, and the above vertical) is compressed once from the -J- input image's 900 and fd