JPS61247183A - Method for reducing coded sample number - Google Patents

Abstract

PURPOSE:To decreas the coded sample number by dividing an input picture into plural blocks, detecting the change in the horizontal and vertical signal levels at each block and controlling the sampling mode based on the result of detection. CONSTITUTION:A block forming circuit 2 forms blocks comprising vertical (m) lines and horizontal (n) samples from the inputted television picture. A block characteristic detection circuit 3 detects the horizontal and vertical signal level change characteristics for each block. A pre-filter circuit 4 smoothes the signal level change in response to the block characteristic ahead the next sub- sampling processing. The sub-sampling circuit 5 varies the coded sample number and sample point arrangement in response to the block characteristic. thus, the coded sample number is reduced while the deterioration of picture quality is suppressed.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a method for reducing the number of encoded samples in a system for encoding image signals such as television signals. It can be effectively used for band compression encoding for digital transmission, digital storage of still image signals, etc.

(Prior Art) A method called subsampling has been known as a method for reducing the number of encoded samples in band compression encoding. (For example, ``Episode, Hashimoto.

Yasuda: Inter-frame coding method using differential amplitude suppression method, IEICE Transactions Vol. 1.60-A, Nα3, 1977.
3J).

(Problems to be solved by the invention) Figure 1 is an explanatory diagram of conventional subsampling (1);
The figures are explanatory diagrams of conventional sub-sampling (2), and in both figures, 0 indicates the encoded sample point and X indicates the sample point to be thinned out.

In this conventional technology, in order to suppress the amount of encoded information, for example, the number of samples in the horizontal direction on the screen is thinned out to 1/2 as shown in FIG. 1, and the number of samples in the vertical direction as shown in FIG. The number of samples is thinned out by half (in this case, it is sometimes called surline) to reduce the number of samples, but this sample number reduction process is not related to the local characteristics of the image. I'm going to

For this reason, in the case of Figure 1, aliasing distortion occurs in parts of the screen where there are many horizontal signal level changes (parts with many horizontal high frequency components), and in the case of Figure 2, aliasing distortion occurs in the parts of the screen where the signal level changes in the horizontal direction. There is a problem in that aliasing distortion occurs in areas where there are many level changes (areas where there are many vertical high frequency components).

Furthermore, in order to prevent this aliasing distortion, there is a method of performing band limitation by pre-filtering prior to subsampling, but in this case, there is a problem that the resolution decreases because high frequency components are lost.

(Means for Solving the Problems) The present invention provides a new method for reducing the number of samples that solves the problem of image quality deterioration caused by conventional subsampling as described above.

The present invention divides an image to be encoded into a plurality of blocks, detects the signal level change characteristics in the horizontal and vertical directions within each block, and based on the detection results, the number of samples to be encoded for each block. The most important feature is that the sample point arrangement is variable.

In the conventional technology, subsampling was performed regardless of the local characteristics of the image, whereas in the present invention, the sampling mode (number of encoded samples and sample point arrangement) is adaptively determined for each block based on the local characteristics of the image. ) is variable.

(Function) In the present invention, the input image is divided into multiple blocks, the horizontal and vertical signal level change characteristics are detected for each block, and the number of encoded samples and sample point arrangement are controlled based on the detection results. Therefore, it is possible to prevent aliasing distortion and a decrease in resolution, which are problems with conventional subsampling, and to reduce the number of encoded samples while suppressing deterioration in the quality of one image.

(Embodiment) FIG. 3 is a block diagram of an embodiment for explaining the present invention in detail, and shows the configuration of a television signal encoding device (transmission side).

1 is an input signal terminal, 2 is a block forming circuit.

3, the block characteristic detection circuit 54 is a pre-filter circuit;
6 is a sub-sampling circuit, 6 is an encoding circuit, 7 is a transmission buffer memory circuit, and 8 is a transmission output signal terminal.

A television signal is input from the input signal terminal 1.

It is assumed that this input signal is in a digital format that has already been A/D converted.

The block forming circuit 2 forms a block consisting of m lines in the vertical direction and n samples in the horizontal direction from the input television image.

FIG. 4 shows an example of a block configuration.

In this example, m=4. With n=4, 4×4=16 samples constitute one block, and the image is divided into a plurality of such blocks.

The block characteristic detection circuit 3 detects local characteristics (block characteristics) of the image corresponding to each block.

This will be explained based on FIGS. 5 and 6.

FIG. 5 is an explanatory diagram of horizontal block characteristic detection.

First, as shown in Fig. 5, each difference d+'' (i = 1 to 12) between adjacent samples in the horizontal direction within a block is determined, and the absolute value of each difference 1di''l is calculated as a predetermined threshold value. The number of occurrences of the difference such that bil"l≧D is counted and calculated by nN&. From this n accent and a predetermined threshold N, the horizontal nature of this block is determined based on the following.

If n=≧N, the change in signal level in the horizontal direction is large, nB
<If N, the change in signal level in the horizontal direction is small.

Similarly, as shown in FIG. 6, the differences d and v (i = 1 to 12) between vertically adjacent samples within the block are determined.

FIG. 6 is an explanatory diagram of vertical block characteristic detection.

The absolute value of each difference, 1 dlvl, is compared with the threshold value, and I
Count the number of occurrences of differences such that dlvl≧D and calculate this as nv
shall be. From this nv and the threshold N, the vertical properties of this block are determined based on the following.

If ≧N, the change in signal level is large in the vertical direction, n
If v < N, the change in signal level in the vertical direction is small.

Based on the detection results of the horizontal and vertical characteristics described above, the characteristics of each block can be summarized as follows.

■nN≧N ・If the signal level change is large in the horizontal direction: n, ≧N ・If the signal level change is large in the vertical direction ■nlI≧N ・If the signal level change is large in the horizontal direction: n v < N ・Signal level change is small in the vertical direction ■nlI<N ・Signal level change is small in the horizontal direction and: n, ≧N ・Signal level change is large in the vertical direction ■n M< N ・When the signal level change in the horizontal direction is small and : nv<N ・The identification information representing ■, ■, ■ of the block characteristics where the signal level change in the vertical direction is small or more is the pre-filter circuit 4 and the sub sampling circuit 5
sent to.

Prior to the next sub-sampling process, the pre-filter circuit 4 performs filtering process to smooth signal level changes according to the block characteristics. The sub-sampling circuit 5 performs processing to reduce the number of encoded samples according to the block characteristics.

The prefiltering and subsampling methods will be explained below based on FIG. 7.

FIG. 7 is an explanatory diagram of subsampling that best represents the features of the present invention.

In FIG. 7, (1) is a case where the signal level change is large in both the horizontal and vertical directions.

(2) is a case where the signal level change is large in the horizontal direction, and the signal level change is small in the vertical direction.

In (3), the signal level change is small in the horizontal direction.

When the signal level change is large in the vertical direction.

(4) shows a case where the signal level change is small in both the horizontal and vertical directions.

When the block characteristic is (■), high frequency components are large in both the horizontal and vertical directions, so if filtering processing is performed in the horizontal or vertical direction, the resolution of the image decreases and the image quality deteriorates.

Furthermore, if horizontal or vertical subsampling is performed without filtering, image quality will deteriorate due to aliasing distortion.

Therefore, in this case, prefiltering and subsampling are not performed as shown in FIG. 7(1).

When the block characteristic is ■, the high frequency component in the horizontal direction is large, but the high frequency component in the vertical direction is small.

Therefore, by performing prefiltering and subsampling only in the vertical direction and not performing prefiltering and subsampling in the horizontal direction, image quality deterioration can be suppressed.

Therefore, after performing filtering processing only in the vertical direction, subsampling processing as shown in FIG. 7(2) is performed to reduce the number of samples in the vertical direction to 1/2.

When the block characteristic is ■, the high frequency component in the vertical direction is large, but the high frequency component in the horizontal direction is small.

Therefore, by performing prefiltering and subsampling only in the horizontal direction and not performing prefiltering and subsampling in the vertical direction, image quality deterioration can be suppressed.

Therefore, after performing filtering processing only in the horizontal direction, subsampling as shown in FIG. 7(3) is performed to reduce the number of samples in the horizontal direction to 1/2.

When the block characteristic is ■, high frequency components are small in both the horizontal and vertical directions.

Therefore, even if prefiltering and subsampling are performed in both the horizontal and vertical directions, almost no image quality deterioration occurs.

Therefore, after performing pre-filtering in both the horizontal and vertical directions, subsampling as shown in FIG. 7(4) is performed to reduce the number of samples to 174.

In this way, by finely controlling the sampling mode (number of encoded samples and sample point arrangement) based on the local properties of the image for each block, it is possible to reduce the number of encoded samples while suppressing image quality deterioration. .

Note that, specifically, for example, the following smoothing filter is used as the pre-filtering.

Example of horizontal prefilter: GM (Z) = 1/4 (z-1+2+z
) Vertical prefilter example: Gv(z) = 1/4 (z−” + 2 + z
”) where z: 1 sample period delay operator, z”
: 1 line period delay operator.

The encoding circuit 6 performs band compression encoding processing on the subsampled signal by, for example, known intraframe predictive encoding or interframe predictive encoding.

At this time, the receiving side is informed whether the sampling mode (number of samples to be encoded and sample point arrangement) of the block to be encoded is shown in Figure 7 (1), (2), (3), or (4). A mode control code for notification is assigned to each block.

Since this embodiment has four sampling modes, the control code is 2 bits/block.

In the buffer memory circuit 7, the code strings outputted from the encoding circuit 6 at non-uniform time intervals are once stored, read out at uniform time intervals, and sent out via the transmission output signal terminal 8.

Note that the block configuration (4x4=
16 samples/block), block characteristic detection (horizontal,
(Detection based on the size of the difference between adjacent samples in each vertical direction)
, the arrangement of sample points, the configuration of the pre-filter, etc. are merely examples of implementation, and it goes without saying that the present invention can be implemented using other methods.

As described above, in the present invention, an input image is divided into a plurality of blocks, and the horizontal and vertical signal level change characteristics are detected for each block.

Since the number of encoded samples and sample point arrangement are controlled based on this detection result, it is possible to prevent aliasing distortion and decrease in resolution, which were problems with conventional subsampling, and to suppress deterioration of image quality while encoding. This makes it possible to reduce the number of samples. Therefore, the present invention is extremely effective when applied to band compression encoding of images that require high quality characteristics.

[Brief explanation of the drawing]

FIG. 1 is an explanatory diagram of conventional subsampling (1). Figure 2 is an explanatory diagram of conventional subsampling (2);
FIG. 4 is a block diagram of an embodiment for explaining the present invention in detail, FIG. 4 is a diagram showing an example of a block configuration, and FIG. 5 is an explanatory diagram of horizontal block characteristic detection. FIG. 6 is an explanatory diagram of vertical block characteristic detection, and FIG. 7 is an explanatory diagram of subsampling that best represents the features of the present invention. 1... Input signal terminal, 2... Block forming circuit, 3... Block characteristic detection circuit, 4... Pre-filter circuit, 5... Sub-sampling circuit, 6
. . . Encoding circuit, 7 . . . Transmission buffer memory circuit, 8 . . . Transmission output signal terminal. Figure 1 110 × ○xOx○X○ X i wlglol 1 Gundelbo・Fig. 2 1-○○○OQ○0000−− − − XXXXX'XXXXXX −−−−00
000ooooo --- 1-XXXXXXXX'XXX-- --○O○○O○○O○○-- Figure 3 Figure 4 Figure 5 d': ; Water i direction m? ! 3Hy7', L41'! t
#nH7ldj'l≧D k'j! ([EI East Fig. 6 ny month d't1 ≧ Dk hrrHt No. 7

Claims (1)

[Claims] In the image signal encoding method, the image is divided into multiple blocks consisting of a fixed number of samples, and the change characteristics of the signal level in the horizontal and vertical directions are detected for each block.
A method for reducing the number of encoded samples characterized by varying the number of encoded samples and sample point arrangement for each block based on this detection result.