JPS58189762A - Picture magnifying and reducing system - Google Patents

Abstract

PURPOSE:To perform a magnifying/reducing process of pictures at a high speed, by loading initially the basic columns to the rotary shift registers and shifting with rotation these basic columns with a timing pulse to extract the information. CONSTITUTION:A processor 401 calculates the basic columns of X, DELTAX, Y and DELTAY and then loads initially them to rotary shift registers 413, 414, 410 and 411 respectively. Then those basic columns are successively shifted with rotation by a timing pulse. An interpolating calculation is carried out on the basis of the overflowed latest four picture elements to extract the information. As a result, the parallel processing is possible to perform a magnifying/reducing of pictures at a high speed.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a digital image scaling method, and in particular,
This invention relates to a method suitable for rapidly obtaining high-quality enlarged/reduced images at arbitrary magnifications.

The entire principle of enlarging and reducing digital images will be explained with reference to FIG. In the figure, (a) is an example of an original image, and (b) is an example of an enlarged/reduced image, which is an image obtained by enlarging the 1 quintillion image (a) by 372 times in the vertical direction and 372 times in the horizontal direction. In the same figure (C1), the enlarged image (b) is mapped onto the original image 1# (Shiage).The enlargement/reduction process is to obtain an enlarged/reduced image t(b)k of the original image (a). , this is the lattice point qll + q,, l... of (b) in the same figure (C).
・・・・・・．

The values of q□,...
pst t・・・・・・・・・l”11+・・・・・・
・It comes down to interpolating based on @kai.

Conventionally, various methods have been proposed for the interpolation rules.

For example, the SPC method, the logical sum method, and the 9-division method.

Projection methods and the like are well known. (See Proceedings of the 20th National Conference of the Information Processing Society of Japan in 1974, pp. 463-464, etc.).

Although the interpolation rules themselves are different for each of the above methods, the input information for the interpolation calculation is common to each method.

Hereinafter, this human power information will be explained in detail with reference to FIG. 2. q
, 1. is the interpolation point, P Xm @ Yta is the interpolation point q19
．． A point on the left among the 4 points near , and an interpolation point q1m on Xfi H
B's original image grid internal address, '/a is the same vertical address,
A is the horizontal length of the original image grid, B is the same vertical length, a is the horizontal length of the enlarged/reduced image grid, and b is the same vertical length. Among these, A, B
is a constant for the device, and a and b are constants for the same magnification.

Input h required to obtain the value of interpolation point Qm, ++!
In the r report, X a + Y'', 4 neighboring points P Xs,
Yw HPXmu + Ya I PXm*+ +Y
+, +1, P Xn HYm+t.

Values in FIG. 2, x, y+++, A, f3. B.

b is a discrete quantity. This is because there is no problem in practical use even if the scaling factor is limited to rational numbers. From this @condition, the following is simpler, so x, y, .

It is assumed that A, B, a, and b take integer values.

In the conventional enlarging/reducing apparatus, the input information necessary for the interpolation described above is obtained by the procedure shown in FIG.

(For example, see Proceedings of the 22nd All Four Conferences of the Information Processing Society of Japan in 1981, 1) I), 73-74).

Us 2XIl @A+X1l V,=Y--B+Y.

It is defined as In the figure 301, VO = 0, 302, y + m = V, mod B, 303, Y, = V, /B, 504, Uo 20, 305, X m = TJ, mod A306, 1
Ding, X, = U, /A 307 detonyl engineering @ 4A, P Xt+ HYm
g PXm+l HYa tP 'Xa
+l I Yawl t tJ XIII Ya
Load the 1st shift of wl from the 1st speed memory. At step 308, interpolation calculation is performed using the results of steps 302, 305, and 307 as all the input information.

In 309, I to + = TJ a + a r to n +1 311 to■□+ =V-+ b m=m+ 1 What is gold? However, in the above step j, mod is an operation for obtaining the remainder of division and the quotient of division by /.

It has been confirmed that if the above procedure is followed, it takes about 1 second to obtain a scaled image consisting of 10'' points.
This corresponds to the fact that it takes about 4 seconds to obtain an enlarged/reduced image corresponding to the size of an A4 paper with a scanning density of 8 lines/listening Fakunmi+). In the above procedure, actually 305, 306 and 3
07,308 can be calculated as 1j. At this time, it is the passes of processes 306 and 307 that govern the processing time.

An object of the present invention is to speed up the processing of enlarged/reduced images by various conventional methods by speeding up all passes of the processes 306 and 307.

In this invention, the X m t yo column and the X a
All of the above objectives have been achieved by utilizing the fact that the columns +I X m + Y m+ and Y m are synchronous columns that repeat base irregularities in a fixed length period. Specifically,
This is achieved by initially loading each basic column into a rotating shift register, rotating and shifting it by a timing pulse to take out the information.

The self-circumferential toughness will be explained below. First, I will clarify about Xl. X, as mentioned above, is xll”U++
It can be defined as rnod A IJ,, =TJ, +a. Regarding this and Nendo's 11, X 1
14A ” [J @+A mod A:=([]a
+Asa) mod A=U, mod A;X1 holds true. Therefore, Xl(g) is a periodic sequence whose period is at most.
It is A-1. y Yu also does exactly the same thing, and the period is still B
In period 1, the basic sequence is yo + Y+ +...
...It can be proven that it is eYR-s. Then X, ya...
-Visiting X. As mentioned above, X can be defined as A @Xa ” [1a X□, -Xn=ΔX.

Then, for any n, A@ΔX e+A ” a (XIl+A+I
m+mu)"a(Xm+I X11) = A-ΔX.

holds true. Therefore, ΔXll+mu=ΔX11 holds true. Therefore, ΔX,=X, , -X, is a periodic sequence with a period of at most A. The basic column is ΔXo.

Δx1. . . . ΔX is 1. ΔY *
Similarly, = Y m+1-Y is a periodic sequence with a period of at most B, and the fundamental sequences are ΔYo, ΔY□,...
, ΔYm-1.

Next, one embodiment of the present invention will be described in detail. FIG. 4 shows the overall configuration of the present invention. 401 is a processing device such as a microcomputer 402 is a processing device [original l [! 11g
A parallel-to-serial converter (hereinafter abbreviated as P/S converter) 403 converts one data into a serial format, and a demultiplexer 404 selects input lines from a buffer memory 4040.
405 is a multiplexer that selects the output line of the buffer memory 404; 406 is a multiplexer 405 that inputs the values of neighboring pixels of the interpolation point, and rotates shift registers 413 and 414; Interpolation #'F calculator that manually scales the grid address xty'r and outputs the entire value of one pixel (7), 407. A serial-to-parallel converter (hereinafter referred to as S/
Abbreviated as a P converter, 408 is a P/S from the processing device 401.
A read/write controller (hereinafter abbreviated as R,/W controller) 4 that controls the transfer to the converter 402 and the transfer from the S/P converter 407 to the processing device ff1t401.
09 is a shift controller that receives the soft count of the buffer memory 404 from the rotating shift register 411 and outputs this (b) number of shift pulses; 410 is a processing unit to be loaded into the buffer memory 404; 111 for each line interpolated
Rotating shift register for next output, 411-J Rotating shift register for outputting the 1111th time every k1 pixel interpolations of the shift count of the buffer memory 404, 412
is a divider that counts interpolation of one pixel for one line, 41
3iu, a rotating shift register that sequentially outputs the X coordinate of the address in the grid of the interpolation point every time one pixel is interpolated, 4
Reference numeral 14 denotes a rotating shift register that outputs the 11th order every time the y-coordinate of the address in the grid of the interpolation point is interpolated for one line. Correlating this diagram with the symbols in Figure 2, x is the output of the rotating shift register 413, and y is the output of the rotating shift register 414.
The output, X ll + I X *=ΔX, is the rotating shift register 411 (7) output, Y,, 1-
=: ΔY, rso ---1i--Output of shift register 410, 1 pulse every time n counts up is input to registers 411 and 413, 1 pulse every time m counts up -h m2 Manual power of registers 410 and 414, P Xa+1 + Y” and P Xm+@
e Yawl becomes the output of multiplexer 405.

In particular, in order to find the neighboring pixels of the interpolation point using ΔX,'(<, the backup memory 404 inputs the shift register.

Next, the operation of this embodiment will be explained with reference to FIG. 50
1 is the scaling factor, lateral force direction A/a.

Frog Mando H/b (However, A=i:! is device specific)
t and l, the processing unit calculates the basic columns of X, Δx, y, and ΔY, and each rotate shift register 410° 411 .
413, 414. At 502, load the first two lines of the IMl town name into the sofa memory 404VC. At 503, load the light, and based on ΔYt, calculate all addresses of the next two lines of the original image. At 504, load the previous two lines of the original dL : Based on the address 503, the next two lines are loaded into the empty line of buffer memory 404.

505.507 The content of the two shift registers being processed in the buffer memory is 1 bit/ft.
'c2 image--Import into the J all-nema calculator 406. In step 506, it is determined whether the software count in step 507 is ΔX11 times. In 508, X is ΔX in 509.
*obtain. At step 510, interpolation calculations are performed based on i''J, xa HYll, and the latest four pixels left over from the buffer memory. At step 511, all determinations of the interpolation stitches for one row are performed.

At 512, y is rotated and shifted by nAYt at 513 to obtain Y11+11ΔY1, respectively.

In step 514, a determination is made as to whether the interpolation for 1lIfI images has ended.

In this embodiment, processes 506 and 50 in FIG.
7,508,509 Processing process 512g513 Processing process 503 and 504,505-513 perform parallel processing. Among these, the total processing time of buses 505 to 513 is equal to the shift time of the number of image pixels of the enlargement/reduction Ii of the shift register. This corresponds to the fact that it takes about 0.6 seconds to obtain an A4 size enlarged/reduced image with a pixel density of 8 lines/size using A, 2827 as a shift register. the other 503 and 50
The total processing time of the 4 bus is equal to the product of the total number of bytes of the original and scaled images times the transmission speed of the bus. This q, 1
This corresponds to the fact that it takes about 1 second to obtain the A4 size image using an M-byte 7-second bus.

As explained above, according to the present invention, by replacing the processes of 305, 306, 307 and 309 in FIG. There is an advantage that it is possible to enlarge or reduce the A4 size image in about 1 second.

[Brief explanation of drawings]

Figures 1 (a), (b), and (C) and Figure 2 are illustrations of image enlargement/reduction, and Figure 3 shows the conventional processing method 11.
1, FIG. 4 is a flowchart showing the configuration of an embodiment of the present invention, and FIG. 5 is a flowchart showing the processing procedure according to the present invention. Whisper 1 Figure (C) %Z Mouth Figure K Kuto

Claims (1)

[Claims] 1. A buffer memory for buffering multiple lines of image data, a controller for controlling input/output of data to the buffer memory, a controller for controlling shifting of the buffer memory, and storage of interpolation address information. An image enlargement/reduction method characterized by having four rotating shift registers.