JPS6126128A - Displacing method of two-dimensional data - Google Patents

Abstract

PURPOSE:To obtain a method for displacing two-dimensional data rapidly by dividing the two-dimensional data into small blocks and executing the displacement of the two-dimensional data in each small block and between blocks independently to displace the whole data. CONSTITUTION:A small block a22 included in undisplaced data 21 stored in a disc is read out to a high speed main storage device 23 on the basis of two- dimensional data displacing algorithm. Local displacement (g) of the block (a) is executed in the main storage device 23, and then the displaced result is stored in a position 25 corresponding to the displaced data 24, so that the data in each small divided block are stored in a continuous addresses of the disc. Said processing is applied to all the blocks of the two-dimensional data 21 to displace the whole two-dimensional data 21. Since the data transfer to the disc is executed in each block stored in a continuous address, high speed processing can be attained.

Description

[Detailed Description of the Invention] [Field of Application of the Invention] The present invention relates to a two-dimensional data transposition method, and particularly a two-dimensional data transposition method that enables high-speed transposition with a small main storage capacity.
Concerning the transposition method of dimensional data.

[Background of the invention]

The E klundh method has been known as a method for transposing two-dimensional data with a small main storage capacity. (J. O. Eklundh, “AF ast
ComputerMethod for Matr
ix Transposing”, IE'Trans
, Comput, , C-21, p801-803.19
72) However, in E klundh's method, transposition is performed by repeating reading and writing of two-dimensional data stored in an external storage device in the column direction, but in order to perform one transposition process, each element of the two-dimensional data must be are read and written multiple times, so the number of data transfers increases, resulting in longer transposition times.
The drawback is that it can only be applied to dimensional data.

[Purpose of the invention]

SUMMARY OF THE INVENTION An object of the present invention is to provide a method that eliminates the above-mentioned drawbacks of the conventional method and executes transposition of two-dimensional data at high speed.

[Summary of the invention]

In order to achieve the above object, the present invention is characterized in that two-dimensional data is divided into small blocks, and the entire transposition is performed separately into transposition within each small block and transposition between blocks. Also, when storing two-dimensional data in an external storage device, by providing free space at the boundaries of the blocks,
It is characterized by shortening the processing time required for storage.

[Embodiments of the invention]

Hereinafter, one embodiment of the present invention will be described as a synthetic aperture radar (S ynt
hetic Aperture Radar; hereinafter SA
The image reproduction processing system (abbreviated as R) will be explained as an example.

FIG. 1 shows the flow of SAR image reproduction processing.

The details of SAR image reproduction processing are conventionally known (for example, J.R., Bennett, I., G.C.
Cumm1n.

“Digital S ARImage Format
im ofAirborne and 5ateli
te Re5ults”, Proe.

13th Int, Symp, Remote Sens
ing of Enviro-nment, vo Q
I, 337-360.1979) Therefore, here, the SAR image reproduction processing related to the following description will be described as two one-dimensional processing, first in the horizontal direction and then in the vertical direction.

The flow of processing will be explained below according to FIG.

Raw SAR data to be processed is read line by line from a magnetic tape 1, and after horizontal one-dimensional processing 2 called range compression, intermediate results are stored in a file 3 on a disk, which is an external storage device.

After performing this processing on all lines, in preparation for azimuth direction FFT processing 6, which is one-dimensional processing in the vertical direction, transpose processing 4 of the two-dimensional data is performed on file 3, and the result is Store in file 5.

Disk devices use random access, but when reading two-dimensional data in the vertical direction in which horizontal data is stored in consecutive addresses, there is a wait for the head to rotate, etc.
It's prohibitively time consuming. Therefore, transposition processing 4 is necessary to shorten the SAR image reproduction processing time.

FIG. 2 shows the state of disk storage for transposing two-dimensional data according to the method of the present invention. The two-dimensional data 11 is divided into small blocks 12, and the data of each block is stored at consecutive addresses on the disk.

FIG. 3 shows a transposition algorithm for two-dimensional data according to the method of the present invention. A small block 22 with data 21 before transposition stored on the disk is read out to the high-speed main memory 23 using the method shown in FIG. 2, local transposition within the block is performed on the main memory 23, and the result is transposed. It is stored in the corresponding position 25 of the subsequent data 24 in the format shown in FIG. By performing the process shown in FIG. 3 on all blocks of the two-dimensional data 21, the entire two-dimensional data 21 is transposed. In the process shown in FIG. 3, data transfer to and from the disk is performed in units of blocks stored at consecutive addresses, so high-speed processing is possible.

In the storage format shown in Figure 2 (hereinafter referred to as the "block format"), data transfer in blocks during transposition processing can be performed at high speed, but data transfer in line units, which is required for processing before and after transposition processing, is generally not possible. It can only be done at low speed. This is because in the block format, data within one line is not at consecutive addresses, so it takes time for the disk to rotate.

An example is shown in FIG. One line of data 32 with two-dimensional data 31 has a logical address of 33.33' on the disk.
, 33' and stored.

Assuming that the size of the small block 34 is 64 pixels vertically, 64 pixels vertically, 8 bytes per pixel, 1 track of the disk = 64 sectors, and 1 sector = 256 bytes, the arrangement of 1 line of data 32 in the disk is as follows: The arrangement will look like the arrangement 36 in the figure. With this arrangement, only 64 pixels of data can be transferred during one revolution of the disk in the rotational direction 35, so line-by-line data transfer is extremely slow.

FIG. 5 shows a disk storage method according to the present invention. Between the data of each block, dummy data 41.41',
By inserting 41',...

The data is arranged in the disk as shown in the arrangement 42 in FIG. 5, and the rotation waiting time is greatly reduced, making it possible to perform line-by-line data at high speed. Dummy data 41.41
Even if ', 41', .

As described above, according to this embodiment, in SAR image reproduction processing, data transfer between the disk and the transposition processing in the transposition processing and the processing before and after the transposition processing can be performed at high speed, and the overall processing time can be shortened.

〔Effect of the invention〕

According to the present invention, as described above, two-dimensional data is divided into blocks and stored on a disk, which is an external storage device,
By providing free space at the boundaries of each block, it is possible to reduce the rotation wait time during line-by-line data transfer, which has the effect of allowing general rectangular two-dimensional data to be transposed at high speed with less main memory. be.

[Brief explanation of the drawing]

Figure 1 is a flowchart of SAR image reproduction processing, Figure 2 is a diagram showing the storage format of two-dimensional data in an external storage device, Figure 3 is a diagram showing the transposition algorithm, and Figures 4 and 5 are inside the disk. FIG. 2 is a diagram showing data arrangement of FIG. 21... Two-dimensional data before transposition, 23... Main memory, 24... Two-dimensional data after transposition, 41.41'
. 41#... Dummy data area, 42... 1st line data Figure 3 ¥35 Figure

Claims (1)

[Claims] 1. When transposing two-dimensional data using a device consisting of an arithmetic unit, a main storage device, and an external storage device, the two-dimensional data is divided into blocks and stored in the external storage device. A two-dimensional data transposition method characterized by: 2. The two-dimensional data transposition method according to claim 1, wherein a free area is provided at the boundary of each block in the external storage device to store the two-dimensional data.