JPH0290375A - Arithmetic processing system for picture data - Google Patents

Abstract

PURPOSE:To shorten the arithmetic processing time by reading only the new original picture element data out of a video memory among plural original picture element data used for calculation of a certain process picture element data and writing the read-out picture element data into a high-speed arithmetic access memory. CONSTITUTION:The video signals received from an input terminal 9 are written into a video memory 6 via an A/D converter 10. Then the video signals are read out of the memory 6 are projected on a monitor picture receiver 8 via a D/A converter 7. The still picture data is written into the memory 6 and magnified or reduced under the control of a computer 1 via a buffer memory 14 and a digital signal processing circuit 15. Then the magnified or reduced picture data is monitored by the receiver 8 and also inputted to a communication process circuit/interface 12 and sent to another still picture communication terminal via a transmission line 13 after a communication process. The process circuit 12 reads the new original picture element data out of the memory 6 except those original picture element data used for calculation of the process picture element data preceding the previous one. The picture element data read out of the memory 6 is written into the memory 14 and undergoes an arithmetic process.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an image data arithmetic processing method suitable for application to a still image processing device of a still image communication device.

[Summary of the invention]

The present invention reads out a plurality of pieces of original pixel data from a video memory in which original pixel data constituting original still image data is stored, and writes the plural pieces of original pixel data to a high-speed access memory for calculation. Processing pixel data at virtual positions near the positions of multiple original pixel data on the screen are calculated by using the original pixel data of In an image data arithmetic processing method that obtains processed still image data corresponding to original still data by repeating the process, among a plurality of pieces of original pixel data used to calculate certain processed pixel data, the previously processed pixel data is By reading only the new original pixel data excluding the original pixel data used for calculation from the video memory and writing it to the high-speed access memory for calculation, it is possible to obtain processed still image data from original still image data. This is intended to shorten processing time.

[Conventional technology]

For example, a rectangle A, as shown in FIG. 5A, is stored in the video memory.
Original still image (natural image) with outlines of B, c, and D.
When data is stored, consider a case where, for example, the original still image is enlarged, and enlarged still image data having the outline of a rectangle A2B2C2D2 is to be stored at an arbitrary position in the video memory. At this time, the original still image data stored in the video memory is
As shown in , just\separate memory (buffer memory)
After that, the original still image data is enlarged, and the enlarged still image data is stored in an arbitrary position of the original video memory in place of the original still image data, as shown in FIG. 5C. I will have to do it again.

In this case, if each side of rectangle A2B2C2D2 is twice the length of the corresponding side of rectangle A+B+C1D1, then in order to make the resolution of rectangle A2B2C2D2 the same as the resolution of the rectangle, The number of pixels in A2B2C2D2 must be equal to the square of the number of pixels in the rectangle A+ B+ C+ Dl.

Therefore, the pixel data of rectangle A282C2D2 is obtained by interpolation processing from the pixel data of rectangles A, B, C, and D.
Pixels within rectangle A2B2C2D2 must be formed. Also, contrary to the above, from the original still image data having the outline of rectangle A2B2C2D2, rectangle A+ B+ C+ D
When trying to obtain reduced processing still data having a contour of
It is necessary to obtain each pixel data within.

Next, a case will be described in which enlarged still image data is obtained from original still image data by spline interpolation.

First, spline interpolation will be explained in one dimension with reference to FIGS. 6 and 7. The feature of spline interpolation is that the curve obtained by interpolation passes through all sample points,
There are no discontinuities, the calculation time is short, the information band is increased compared to complete convolution, and the curve obtained by interpolation is a cubic curve.

The spline function is a sum expression of interval functions expressed by cubic functions that differ for each two-sample point interval.

That is, as shown in FIG. 6, the interval (X I Xo≦X<X
l ), (x-1, f-1), (Xo, fo)
, (X+, fI), and (X2, f2). (x)=a. x3+b.

It is a function that is equal to x” + cm x 10d.

Its definition is expressed as follows (see Figure 7)
.

f (x)=a-6x3+bnx2+cnx2+d1
1 However, (a fI, b n, c fI, d n)
is given by the following equation (see Figure 7).

f i'l+2= f n (x n+2) f fl+
I = f n (X nn) ffI = f11 (xn
) f fl-1=f Tl (xn-1) Now, the sampling intervals are equal intervals, which are normalized to 1.

Normalize to Furthermore, X'll≦X<Xfl+1
Considering the normalized sentence, intersection = (x-xn) / (xnya, Xn)-x-x-
6, solving the previous function f (x) (-y) yields the following equation.

y-(1/6) (f 'I'12 3 f ffI
1 +3 f n -f 'nt:l 9.3+ (
1/6) X (3f n, l 6 f n+3 f
Tl-1) Cross 2+ (1/6) (f n+2+6
f'n+13 f n-2f n-+), 9. +
(1/6) X(6fn) Rewriting this equation yields the following equation.

3'=ffl-1, f fI, f fl+I, fnya,
)× or more will be expanded to two dimensions and explained with reference to FIG. That is, consider finding the pixel value (pixel data) at point Q. In obtaining the pixel data at point Q, first, gos
Find g+, g2, g3, and on the line connecting each point,
Find the pixel data of point Q.

gm= (fOm, flm, f2 m, f3 m) x Note that (A) is expressed by the following formula, and (At) is the transposed matrix of (A).

Here, Y-(1, y, , y2, y3) From the above, pixel data Q can be obtained by the following equation.

If we set Q= (1, y, y2, y3) (At)×,
Pixel data Q is expressed by the inner product of five matrices as shown in the following equation.

Q='l'・Mut −F・Mu・X Next, pixel data P (xl,
Pixel data Q (X,
y) -Y (y+) ・At-F (xl.

A conventional example of arithmetic processing for determining y+) .mu.X(x+) will be explained with reference to the flowchart of FIG. First, set y = startY (initial value) (step 5T-1). Then x = s
Set tart-X (initial value) (Step 5T2)
. Next, the calculation of -m·x (x) is performed (step 5T-3). Here, t indicates a temporary matrix of (IXn). Next, the calculation t=F (x, y)·t is performed (step 5T-4). Next, the calculation t=mut-t is performed (step 5T-5).

Next, the calculation P(x,y)-so(y)·t is performed (step 5T-6). Next, set x=x+1 (step 5T-7). Next, in step 5T-8, X is en
It is determined whether or not it has reached d-X (final value), and if No, the process returns to step 5T-2, and if YES, the process proceeds to step 5T-9. In step 5T-9, y=y+
Put it as t. Next, in step 5T-10, y is end-
It is determined whether or not it has reached Y (final value). If NO, the process returns to step 5T-2; if YES, the process ends.

The conventional arithmetic processing shown in FIG. 9 has a disadvantage in that the processing time is extremely long.

Next, a conventional example will be explained with reference to FIG.

For example, in FIG. 11, original pixel data po, p+, P2, . . . , Pl4 constituting the original still image data
When trying to obtain enlarged pixel data qos q+ , q2, ... q+s, etc. that constitute enlarged still image data from Data Po, Pl, P2, P5, P6
, Pl, PH0s P++, Pl2. In addition, the enlarged pixel data q2 is
3×3 original pixel data P. , Pl, P2, P5, P6
, Pl, PIO% pH, and Pl2. In this way, the original pixel data required for this operation is repeatedly retrieved from video memory.

This will be explained with reference to the flowchart in FIG. The 3x3 original pixel data from the video memory (video RAM) where the original still image data is stored is transferred to the work R.
Transfer to AM (step 5T-1). Next, work R
A single piece of enlarged pixel data is calculated using 3×3 pixel data stored in the AM. next,
Step ST” to determine whether the predetermined calculation has been completed or not.
-3), if No, return to step 5T-1, YE
If it is S, it ends. The enlarged pixel data q thus calculated. , (11, (12,...) as the first
As shown in Figure 2, instead of the original original pixel data, it is stored in the original video RAM as QO, Ql, Q2, .

As described above, it takes time to repeatedly read the original pixel data from the video R'AM, leading to an increase in processing time.

[Problem to be solved by the invention]

In view of this, the present invention proposes an image data arithmetic processing method that can shorten the arithmetic processing time.

[Means to solve the problem]

The present invention reads out a plurality of original pixel data from a video memory in which original pixel data constituting original still image data is recorded, writes them to a high-speed access memory for calculation, and stores the data in the high-speed access memory for calculation. Processing pixel data at a virtual position near the position of the plurality of original pixel data on the screen is calculated by using a plurality of original pixel data, and this is calculated as processing pixel data at a virtual position adjacent to the screen in sequence. In a still image arithmetic processing method that obtains processed still image data corresponding to the original still data by repeating each processing, among the plurality of original pixel data used to calculate a certain processed pixel data, the previously processed pixel data Only the new original pixel data excluding the original pixel data used for calculation of
The data is read from the video memory and written to the high-speed access memory for calculation.

[Effect]

According to the present invention, a plurality of pieces of original pixel data are read out from a video memory in which original pixel data constituting original still image data is stored and written to a high-speed access memory for calculation, and the data is stored in the high-speed access memory for calculation. Processing pixel data at a virtual position near the position of the plurality of original pixel data on the screen is calculated by using a plurality of original pixel data, and this is sequentially applied to processing pixels at adjacent virtual positions on the screen. In a still image arithmetic processing method that obtains processed still image data corresponding to the original still data by repeating each data, among the plurality of original pixel data used to calculate a certain processed pixel data, the previously processed pixel Only new original pixel data excluding the original pixel data used for data calculation is read from the video memory and written to the high-speed access memory for calculation.

〔Example〕

Embodiments of the present invention will be described in detail below with reference to the drawings. First, a still image processing device of a still image communication terminal to which the present invention is applied will be explained with reference to FIG.

(1) is a computer (microcomputer), C
It is composed of a PU (central processing unit) (2), a ROM (3), and a RAM (4-). (5) is the CPU (
2) bus (consisting of data bus, address bus, control bus, etc.), ROM (3) and RAM (4,) are connected to this bus (5). This computer (1
) controls each part of this still image communication terminal. (13)
is a transmission line, which can be wireless or wired, but wired transmission lines include l5DN (Ingrated Services Digital Network), high-speed digital line, analog telephone line, and DDX (Digital Data Exchange Network). (There are two types of this, DDXC and DDXP), a dedicated line, etc.

42) is a communication processing circuit and an interface connected between this transmission line (13) and the lotus (5) according to the protocol and transmission speed of the still image signal of the transmission line (13); The communication processing circuit performs transmission processing and reception processing, such as encoding and modulation for transmission, and decoding and demodulation for reception, respectively.

(6) is a frame memory (video RAM) whose digital video signal input/output terminal and control signal input terminal are connected to the lotus (5), and whose digital video signal input terminal is connected to the A/D converter (10). The digital video signal output terminal is connected to the input terminal of the D/A converter (7), and the other control signal input terminal is connected to the output terminal of the display timing control circuit (11). be done. The writing and reading of this memory (6) is controlled by the computer (1) and the display control circuit (11). Note that this memory (6) includes horizontal and vertical address counters, a memory controller, etc.

(9) is an analog video signal input terminal, and the video signal from this input terminal (9) (video signal from a video camera, VTR, etc.) is supplied to the A/D converter (10) and converted into a digital video signal. After being converted, it is supplied to the memory (6) and written therein.

The digital video signal read out from the memory (6) is
After being supplied to a D/A converter (7) and converted into an analog video signal, it is supplied to a monitor receiver (8) equipped with a cathode ray tube and displayed on the screen of the cathode ray tube.

As described above, the analog video signal supplied to the input terminal (9) is supplied to the A/D converter (10), converted to a digital video signal, and then supplied to the memory (6). Although it is written as image data of a still image, it is stored in the Hasofa memory (14) and the digital signal processing circuit (16).
) is used to enlarge or reduce the image under the control of the computer (1), and then it is monitored by the monitor receiver (8) and is also supplied to the communication processing circuit and interface (12) for communication. After being processed, it is transmitted to other still image communication terminals through the transmission line (line 3).

(14) is a sofa memory whose input/output terminals are connected to the bus (5), and whose writing and reading are controlled by the computer (1).

(15) is a digital signal processing circuit whose input/output terminals are connected to the bus (5), and which is connected to the computer (1).
controlled by Note that (16) is a high-speed access RAM provided in the digital signal processing circuit (15), and its access speed is significantly higher than that of the video RAM (6).

Now, in FIG. 11, P (xy) is
Noting that these are all points that exist in a two-dimensional space where 0≦x<m and 0≦y<n, all X between 0 and m-1 and all points between 0 and n-1 Preprocessing is performed on y to create vector arrays consisting of X(x)-x (x) and Y (y), respectively. (Then, by using these arrays X and Y, the operation of P (x, y) is p (x, y) - Y (y) ・F (x
, y)X (X), the number of product-sum operations is reduced to n3, and the processing time is approximately 1
It only takes seconds. The ratio is 1:4000.

Next, this preprocessing as an embodiment of the present invention will be explained with reference to FIG. First, X=start −
Set it as X (initial value) (Step 5T-1).

Next, in step 5T-2, X (X)-8・x(X)
Perform the calculation. Next, in step 5T-3, X is changed to
Set it as 1. Next, in step ST4, X is set to the final value en
It is determined whether or not d-X has been reached. If NO, return to step 5T-2; if YES, return to step 5T.
-Move to 5.

Next, in step 5T-5, y is set to 5tart-Y (
initial value). Next, in step 5T-6, Y (
y) −y (y) ・Perform the calculation of Mut. Next, in step 5T-7, y=y+1 is set. Next, in step 5T-8, it is determined whether y is end-Y (final value). If NO, the process returns to step 5T-6, and if YES, the process proceeds to RET (return) (step 5T-9).

Next, with reference to FIG. 2, the main processing after the preprocessing will be explained. In step 5T-1, after the preprocessing in FIG. 1 is performed, in step 5T-2, y = start
-Y (initial value). Next, in step 5T-3,
Set x = start-X (initial value).

In step 5T-4, t=F (x, y) ・X (
x) is performed. Here, t indicates a temporary matrix. In step 5T-5, P (x, y) = Y
(y) ・The calculation of t is performed. Next, step 5
At T-6, set x=x+l. Next, step S”17
Then, it is determined whether or not X is end-X (final value). If NO, the process returns to step STi, and if YES, the process moves to step 5T-8. Step 5T-8
Now, let's set y=y+t.

In step 5T-9, it is determined whether y is end-Y (final value), and if NO, step 5T-
Return to step 3, and if YES, end.

Next, the present invention will be described in detail with reference to the flowchart of FIG. First, 3×3 original pixel data (for example, P O-, P 1 , P2, PS, Po in FIG.
・Pl・PIO・pH, Pl2) is transferred to the work RAM (16) of ・DSP (digital signal processing circuit>(15)) (step ST-1).

Next, 3×3 original pixel data P stored in the work RAM (16). , Pl, P2, PS, Po・Pl・
Calculate and output one enlarged pixel data (q+ in Figure 11) by calculating using PIO, pH, and Pl2 (
Step 5T-2). Output enlarged pixel data q
1 is written and stored at a predetermined address of the video RAM (6) via the panzer memory (14), as shown in FIG.

Next, it is determined whether the necessary calculations have been completed, and if YES, the process ends, and if NO, the process moves to step 5T-4.

In step 5T-4, it is determined whether the virtual position of the processed pixel data has moved to the next one, that is, the original pixel data PH, Po, P+
Since it has moved between + and P2, Pl, Pl2, the process moves to step 5T-5.

Po1P5,P. and Pl, Po, P, 1, the original pixel data PO, PS, PIO, P2, Pl,
It is determined whether or not it has moved between Pl2.

If NO, the process moves to step 5T-6, and if YES, the process moves to step 5T-5. In the example of Figure 11,
The virtual position of the processed pixel data moves from q to q2,
Original pixel data PO%P5, PIO, p2, Pl, P
12, the process moves to step 5T-5.

In step ST''-5, 3Xn pieces of original pixel data PH, P2, Po, Pl, pH, P
12 to a predetermined address in the work RAM (16), and transfer the new 3×(3-n) from the video RAM (6).
, here, three original pixel data P3, Pg, Pl3
is transferred to the work RAM (16). After that, the process moves to step 5T-6.

In step 5T-6, 3×3 original pixel data P1, P2, PS,
A calculation is performed using Po, Pl, P8, pH to P12, PI3, and one enlarged pixel data q2 is calculated and output. The output enlarged pixel data q1 is written and stored at a predetermined address in the video RAM (6) via the panzer memory (14), as shown in FIG. 12.

After that, the process moves to step 5T-7.

In step 5T-7, it is determined whether the necessary calculations have been completed. If YES, the process ends; if NO, the process returns to step 5T-4.

In this way, from now on, the enlarged pixel data q3 and q4
・q5, q6, q7, etc. are calculated and output, and the first
As shown in FIG. 2, the data is written and stored at a predetermined address in the video RAM (6).

In the above-described embodiment, the case of enlargement processing was described, but it may also be case of reduction processing, in which case thinning processing is performed on the original pixel data.

〔Effect of the invention〕

According to the image data calculation processing method of the present invention described above, the time for calculating processing pixel data is shortened.

[Brief explanation of drawings]

1, 2, and 3 are flowcharts of an embodiment of the present invention, FIG. 4 is a block diagram of an embodiment of the present invention, FIG. 5 is an explanatory diagram of enlargement processing, and FIG. Figures 7 and 8
The figure is an explanatory diagram of spline interpolation, Figure 9 is a flowchart of a conventional example, Figure 10 is a flowchart of a conventional example, and Figure 11 is a flowchart of a conventional example.
FIG. 12 and FIG. 12 are explanatory diagrams of the enlargement process. (1) is computer, (2) is CPU, (3) is RO
M, (4) is RAM, (6) is video memory (RAM
), (14) is a sofa memory, (15) is a digital signal processor, and (16) is its work RAM.

Claims (1)

[Claims] A plurality of original pixel data are read out from a video memory in which original pixel data constituting original still image data is stored and written to a high-speed access memory for calculation, and the data is stored in the high-speed access memory for calculation. Processing pixel data at a virtual position near the position of the plurality of original pixel data on the screen is calculated by using a plurality of original pixel data, and this is calculated using the plurality of original pixel data at a virtual position adjacent to the position on the screen. In an image data arithmetic processing method that obtains processed still image data corresponding to the above-mentioned original still data by repeating processing for each processed pixel data, one of the plurality of original pixel data used to calculate a certain processed pixel data, the previous An image data operation characterized in that only new original pixel data excluding the original pixel data used for calculation of previously processed pixel data is read from the video memory and written to the operation high speed access memory. Processing method.