JPH04142678A - Image expansion processing method in affine conversion - Google Patents

Abstract

PURPOSE:To reduce the number of times of accessing an image memory and to speed up image expansion processing in affine conversion by storing a frequency of using input image data in a data memory of a pipe line processor. CONSTITUTION:From an input magnification ratio, the number of times of accessing the same address by image memory 1 is calculated and stored in a data memory 34 of a pipe line processor 3. Next, an address of image memory 1 in which output data is to be written is calculated. Further, data is read from image memory 1, and a processing unit 36 converts read data into output data nd is stored in image memory 1. Furthermore, input data read out in advance is used to prepare output data, the average value of the data is used to correct the data repeatedly. With this, the number of times of accessing the image memory is reduced and image expansion processing in affine conversion is speeded up.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an image enlargement processing method in affine transformation of an image processing system, and particularly to an enlargement processing method using a pipeline processor.

[Conventional technology]

In affine transformation, the X and Y coordinates on the input side are determined from the X and Y coordinates on the output side, and the address in the image memory corresponding to the coordinates is accessed.

Particularly in the case of enlargement processing, the same data will be accessed many times depending on the magnification.

For example, 2×2 4-pixel image data 10 (a+++ ax2+ a2t+ a2□
) is doubled in the X direction and doubled in the Y direction, each pixel of the enlarged image data 20 is A s
1= A 12= A 21= A 22= a l1
lA s3= A 14= A 23= A 24=
a 121Aa1=A32=A41=A42=a2! I
A 3G= A 34= A 43” A 44= a
22+, and the same data is used in each of the four locations.

In the conventional image enlargement processing method using affine transformation, the image memory is accessed once each time one output pixel is obtained.

[Problems to be Solved by the Invention] The conventional image enlargement processing method in affine transformation described above has a problem in that when obtaining image data corresponding to each pixel of output data, even in the case of enlargement processing, when using the same input data. However, since the image memory is accessed every time, the processing time increases.

[Means to solve the problem]

The image enlargement processing method in affine transformation of the present invention includes a first step of calculating the number of accesses to the same address in the image memory from a specified enlargement rate, and a pipeline processor a third step of calculating an address in the image memory from which the data should be read; and a fourth step of calculating an address in the image memory to write the generated output image. and a fifth step of reading data from the image memory from the address determined in the third step, and reading the generated output data from the third address in the image memory.
The sixth step is to write to the address obtained in step 1, and the seventh step is to repeat creating output data using only the value obtained in the first step from the read data.
and an eighth step in which the processes from the third step to the seventh step are repeated until all data for each pixel of the output image is read out.

〔Example〕

Next, the present invention will be explained in detail with reference to FIGS. 1 and 2.

Referring to FIG. 2, the information processing system used in one embodiment of the present invention is comprised of an image memory 1, a memory control device 2, and a pipeline processor 3.

The image memory 1 is a memory that stores image data, and the memory control device 2 controls data input/output between the image memory and the pipeline processor.

The pipeline processor 3 includes an input controller 31, a link table 32, and a function table 3.
3. Data memory 34. Queue buffer 35, processing unit 36 and output controller 3
7 and generates output data from input data.

Referring to FIG. 1, one embodiment of the present invention includes a step S1 of calculating the number of accesses to the image memory, a step S2 of storing the value obtained in step S1 in the data memory 34, and an address for writing output data. Step S3 of calculating an address for reading data from the image memory, Step S4 of calculating an address for reading data from the image memory, Step S5 of reading data from the address in the image memory determined in S4, and Step of writing the generated output data to the image memory. The process includes step S6, step S7 in which the read data is repeated a plurality of times to create output data, and step S8 in which steps S3 to S7 are repeated.

First, calculate the number of accesses to the same address in the image memory 1 from the input magnification ratio (step S1).
, this value is stored in the data memory 34 in the pipeline processor 3 (step S2).

Next, calculate the address when writing output data to image memory 1 (step 3), and calculate the address when reading data from image memory 1 (step 4).
do. Then, data is read from the image memory 1 (step 5).

The data read from the image memory 1 is converted into output data by the processing unit 36 via the link table 32, function table 33, data memory 34, and queue buffer 35 in the pipeline processor 3, and written to the image memory. (Step 6).

Conventionally, the data input here is discarded and new data is read out from the image memory 1, but in the case of enlargement, when the same input data is required again, it is obtained in advance in step S2 and stored in the data memory 34. Referring to the stored values, without reading the input data from the image memory 1 again, the previously read input data is used to create output data (step S7), and data 1 is stored in the image memory. Write.

When writing output data using a real number magnification instead of an integer magnification, if the data has already been written to image memory 1, the average value of the data values is taken and the data is corrected (step 8).

The processes from step 3 to step 8 are repeated until all data for each pixel of the output data is read out and the output data is written (step 9).

〔Effect of the invention〕

As explained above, the present invention stores the number of times input image data is used in the data memory in the pipeline processor, thereby reducing the number of accesses to the image memory and speeding up affine transformation enlargement processing. There is an effect.

[Brief explanation of drawings]

FIG. 1 is a flowchart of processing according to an embodiment of the present invention, FIG. 2 is a block diagram of an information processing system used in this embodiment, and FIG. 3 is a diagram for explaining general image enlargement processing. DESCRIPTION OF SYMBOLS 1... Image memory, 2... Memory control device, 3... Pipeline processor, 31... Input controller, 32... Link table, 3
3... Function table, 34... Data memory, 35... Queue buffer, 36... Processing unit, 37... Output controller.

Claims (1)

[Claims] A first step of calculating the number of accesses to the same address in the image memory from a specified enlargement ratio; and storing the value determined in the first step in a data memory in a pipeline processor. a second step of calculating an address in the image memory to read the data from; a fourth step of calculating an address in the image memory to write the generated output image from the image memory. a fifth step of reading data from the address determined in the third step; a sixth step of writing the generated output data to the address determined in the third step in the image memory; The seventh step corrects the data by taking an average when the written images overlap when they are not multiples of an integer, and creates output data by using only the values found in the first step from the read data. and a ninth step in which the processes from the third step to the eighth step are repeated until all data for each pixel of the output image is read out. Method.