JPH0143513B2 - - Google Patents

Description

Detailed Description of the Invention (Industrial Application Field) The present invention provides a method for subsampling and recording, encoding/transmitting/decoding a moving image signal such as a television signal. Therefore, when performing interpolation processing to restore the video signal sampled at the original fundamental sampling frequency with high accuracy,
The present invention relates to a method for performing adaptive processing that takes into account the amount of motion of a moving image signal, and can be applied to, for example, high-efficiency encoding of television signals.

(Prior Art) A video signal such as a television signal is a three-dimensional signal (space+time), and the signal sampled at the fundamental sampling frequency has a huge amount of information. Therefore, the signal can be recorded, encoded, transmitted,
When decoding, you can easily reduce the amount of information by 1/2, 1/3,...
Conventionally, subsampling has been widely used as a method that can reduce the amount of noise.

Various types of subsampling methods, ie, subsampling patterns, have been proposed.

Incidentally, the image signal is a continuous still image signal (frame signal), and each frame is composed of a fixed number of scanning lines. Generally, this scanning does not sequentially scan the image from the top, but first scans the odd scan lines and then scans the even scan lines. That is, one image is divided into two scans and then converted into an image signal (two-field signal), which has the effect of reducing flicker on a CRT.

For this reason, the above-mentioned sub-sampling patterns are also classified based on whether they repeat or alternate on a frame-by-frame basis, but here we will focus on the 1-line PASS type sub-sampling pattern in which the sub-sampling pattern is alternated on a frame-by-frame basis. We will now discuss a method for obtaining an image quality almost equivalent to that of an image signal without subsampling from an image signal obtained by applying this subsampling pattern.

FIG. 2 is a diagram showing this 1-line PASS type subsampling pattern, in which subsampling is performed every other pixel in a certain frame, and the pixels to be subsampled are alternated every scanning line. Furthermore, it is a subsampling pattern in which pixels to be subsampled are alternated in odd frames and even frames, and has the following three characteristics.

(1) As shown in FIG. 3, when restoring the original image signal by vertical interpolation processing, this can be done by using pixels within the same field. Although the image is blurred in the direction of movement due to the integral effect of the image pickup tube, in this case a good interpolated image can be obtained for vertical movement. On the other hand, image quality deterioration occurs due to interpolation for horizontal movement.

(2) As shown in FIG. 4, when restoring the original image signal by horizontal interpolation processing, this can be done by using pixels within the same field. Although the image is blurred in the direction of movement due to the integral effect of the image pickup tube, in this case a good interpolated image can be obtained for horizontal movement. On the other hand, for vertical motion, image quality deterioration occurs due to interpolation.

(3) Since the subsampling pattern is a frame alternation type, the original image signal can be restored by interpolation processing in the time axis direction, as shown in FIG. In this case, when the amount of image movement is close to 0, the original image is faithfully reproduced. On the other hand, when the amount of images is large, image quality deterioration occurs due to specific interpolation.

(Problems to be Solved by the Invention) As mentioned above, although the conventional interpolation methods have the characteristic of restoring good image quality to image signals of specific motion,
General moving images exhibit various movements in the vertical and horizontal directions, and the amount of movement may be large, so when conventional interpolation methods are widely applied to these general moving images, specific deterioration occurs. It has the disadvantage that image quality is significantly degraded.

The present invention was devised in view of the above-mentioned drawbacks of the prior art, and aims to prevent image quality deterioration that occurs when any of these interpolation methods is used fixedly, while taking advantage of the unique features of the above-mentioned interpolation methods. shall be.

(Means for Solving the Problems) A feature of the present invention is that the interpolated values obtained by the above-mentioned various interpolation methods are appropriately adjusted to the interpolated values using information regarding the movement of the image signal around the interpolated pixel. By adding a weight and taking an average value and using this as the pixel value of the interpolated pixel, the purpose is to prevent deterioration of image quality due to interpolation even for image signals with a wide variety of motion amounts. .

(Operation) The adaptive interpolation according to the present invention enables interpolation that does not cause image quality deterioration even when the image moves in various directions. In other words, if the movement in the horizontal direction is large, the interpolation shown in Figure 4 is automatically performed, and if the movement in the vertical direction is large, the interpolation shown in Figure 3 is automatically performed, and for still images, the interpolation shown in Figure 4 is automatically performed. A method is realized in which the image quality, that is, the resolution does not deteriorate even when the 5 interpolations are performed and compared with the original image signal before subsampling.

(Example) The present invention will be described in detail below. As shown in FIG. 6, six pixels A to F are used for interpolation.
Even better interpolation image quality can be obtained by considering pixels other than these six pixels, but in order to simplify the explanation, an interpolation method using these six pixels will be described below.

As mentioned above, when the image moves horizontally due to the integral characteristics of the image pickup tube, the absolute value of the difference value of pixel values |A-B|, |C-D|, |E-F| Of these, |A−B| is the minimum. On the other hand, when the image moves in the vertical direction, |C-D| becomes the minimum, and when the image is a still image, |E-F| becomes the minimum. That is, from this, |A-B|, |C-
Information regarding the moving direction of the image signal can be obtained from the magnitude relationship of D| and |EF|.

On the other hand, the following relational expression can be obtained from a simple image signal model.

P{|A-B|=min(|A-B|, |C-D|, |E-F|)
}~1/A-B/1/A-B+1/C-D+1/E-F
△＝P _h ……(1) P｜｜C-D｜=min (｜A-B｜, ｜C-D｜, ｜E-F｜)
}~1/C-D/1/A-B+1/C-D+1/E-F
△＝P _v ……(2) P｛｜E‐F｜=min(｜A‐B｜, ｜C‐D｜, ｜E‐F｜)
}~1/E-F/1/A-B+1/C-D+1/E-F
1/A-BP _t ...(3) Now, let the interpolation signal be the weighted average of the six pixels A to F, multiply this by the probabilities obtained by equations (1) to (3), and calculate the average. By using this value, an adaptive interpolation value corresponding to the amount of motion of the image signal can be obtained using the following equation. That is, interpolation signal = P _h × (horizontal interpolation signal) + P _v × (vertical interpolation signal) + P _t × (time axis interpolation signal) 1/1/A-B+1/C-D+1/E-F{1/ ｜A-
B｜ △+B/2+1/｜C-D｜ C+D/2+1/｜
E-F｜E+F/2...(4)

Note that the adaptive interpolation coefficient P _h used in the above explanation,
There are many possible ways to give P _v and P _t other than equations (1), (2), and (3). For example, P _h =1-|A-B|/|A-B|+|C-D|+|E-
F|P _v =1−|C−D|/|A−B|+|C−D|+|E−
F | P _t = 1- | E-F | / | A-B | + | C-D | + | E-
F | ...(5) P _h = 1 | A-B | = min ( | A- B |, | C- D |, |
E-F |) 0 Other P _v = 1 | C-D | = min ( | A-B |, | C-D |, |
E-F |) 0 Other P _t = 1 | E-F | = min ( | A-B |, | C-D |, |
E-F |) 0 Others ...(6) is possible. If these adaptive interpolation coefficients are used,
It is possible to reduce the amount of hardware compared to using equations (1), (2), and (3).

FIG. 1 shows an example of a circuit configuration for adaptive interpolation. In FIG. 1, 1 and 2 are frame memories, 3 is a weight calculation unit, 4 is an interpolation value calculation unit, 5, 6, and 7 are multipliers,
8 and 9 are adders.

The operation of this embodiment will be explained below. The subsampled input image signal is stored in frame memory 1,
Of the pixel values A, B, C, D, E, and F that are sequentially stored in frame memory 2 and used for adaptive interpolation, A, B, C, and D are stored in frame memory 1 to E from frame memory 2, and F is stored in frame memory It is read out without being stored and input to the weight calculation section 3 and the interpolation value calculation section 4. The weight calculation section 3 is easily constructed from known circuit elements, and calculates the weight P according to equations (1), (2), (3) or (5), (6).
_h ,
Calculate _Pv , _Pt . On the other hand, if all such calculation results are obtained in advance and stored in the ROM, high-speed processing is possible. The interpolation value calculation unit 4 calculates 1/2(A+B), 1/2(C+D), 1/2(E+) from the input pixel values A to F.
F). The output of this interpolation value calculation unit 4 is multiplied by weights P _h , P _v , P _t in multipliers 5, 6, and 7, respectively, and then summed in adders 8 and 9, resulting in the final adaptive interpolation. This becomes the output.

(Effects of the Invention) As described in detail above, according to the present invention, by performing interpolation that is adapted to the amount of image motion on a subsampled image signal, various motions can be displayed, and the amount of motion can be reduced. The image quality does not deteriorate significantly even with large moving image signals, and the amount of information is reduced to 1/2 when recording, encoding, transmitting, and decoding.
It can be reduced to 2, 1/3, etc., and is effective for high-efficiency encoding in band compression digital transmission of television signals.

[Brief explanation of drawings]

FIG. 1 is an example of a circuit configuration for adaptive interpolation according to the present invention, FIG. 2 is a diagram showing a frame alternating one-line PASS type subsampling pattern to which the present invention is applied, and FIGS. 3, 4, and 5 are for conventional fixed interpolation. FIG. 6 is a diagram showing the spatio-temporal arrangement of pixels used for interpolation. 1, 2; frame memory; 3; weight calculation unit;
4; Interpolated value calculation unit, 5, 6, 7; Multiplier, 8,
9; Adder.

Claims (1)

[Claims] 1. In an image signal interpolation method that restores the original sampled image signal by interpolation from a signal whose information content has been reduced by subsampling the sampled image signal, the pixel values surrounding the interpolated pixel are Find the adaptive interpolation coefficients in the horizontal, vertical, and time axis directions, and the interpolated values in the horizontal, vertical, and time axis directions, and find the sum of the products of the adaptive interpolation coefficients and the interpolated values in each direction. , an adaptive interpolation method for image signals characterized by using this as the interpolated pixel value.