JPH02137489A - Image encoding transmission equipment including motion compensation - Google Patents

Abstract

PURPOSE:To obtain satisfactory visual characteristic in a reproducing image by including an invalid gray level block smoothing means which performs a smoothing processing on a picture element that is an invalid block and whose most approximate block is a gray level block in a smoothing means. CONSTITUTION:Firstly, a loop filter control part 18 inputs most approximate block selection information and valid/invalid information. When it is the valid block, an instruction to turn on a filter processing and an ordinary filter coefficient(smoothing characteristic) K, for example, K=0.7 to a loop filter 17. Also, when it is the invalid block, it is further judged whether or not the most approximate block is the gray level block. The gray level block is selected as the most approximate block, and also, the instruction to turn on an in-loop filter processing for the block judged as invalid is outputted to the loop filter 17. Simultaneously, the coefficient K for the in-loop filter processing is designated at a level lower than an ordinary one(for example, K=0.4).

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an image encoding and transmitting apparatus including motion compensation, and an image encoding and transmitting apparatus for transmitting moving images such as video conferences or video telephones.

[Prior Art] In general, image transmission used for video conferences, video calls, etc. requires a huge amount of image information to be transmitted. Therefore, in terms of line speed and transmission cost, it is necessary to reduce the amount of image information to be transmitted (
A coding transmission method that performs compression) has been put into practical use.

Among these, vector quantization devices that include motion compensation are known as information compression and transmission devices that have a high degree of compression.

A conventional image coding and transmission apparatus including motion compensation will be described below.

Figure 4 shows “Vector quantization method for color images; Murakami (
(2 others)” (Television Society Journal Vol. 1.39°N)
FIG. 3 is a block configuration diagram of a transmitting unit of a coding/transmission apparatus to which the quantization/coding method disclosed in No.

In the figure, (1) is a transmitting section, (2) is a block configuration section, and (3
) is the gray level generator, (4) is the frame memory, (
5) is a read control section, (6) is a write control section, (7) is a distortion calculation section, (8) is a delay circuit, (9) is a nearest block selection section, (10) is a subtracter, (11) is a DC separation normalization calculator, (12) is a valid/invalid discriminator, (13) is a local decoder, (14) is a Doji conversion encoder, (15) is a quantization decoder, (16) is a threshold value controller, (17) is an in-loop filter, and (19) is a transmission line encoder.

Next, the operation will be explained.

First, the block configuration unit (2) digitizes an image signal input from a television camera (not shown) or the like using an A/D conversion unit (not shown), and puts the digitized image signal series in rask order. input.

Then, the block configuration unit (2) blocks the image signal sequence into adjacent pixels on the image one by one, converts the arrangement of the signal sequence, and generates an input block signal consisting of K pixel level groups. Output.

On the other hand, the readout control unit (5) reads a block signal at a position corresponding to the input block signal on the screen and a block signal located in the vicinity thereof from the frame memory (4) that stores the decoded reproduction signal of the previous frame. It is read out in synchronization with the input block signal as a motion compensation block signal series.

Further, following the motion compensation block signal series, the gray level generating section (3) generates a block signal (hereinafter referred to as a gray level block signal) in which the pixel level of each pixel is a constant value, in synchronization with the input block signal. ) is output.

Then, the distortion calculation unit (7) receives the input block signal,
A motion compensation block signal series and a gray level block signal series are read synchronously, and distortion values (absolute value distortion values,
Euclidean distortion value, etc.). At this time, assuming that there are a total of n blocks in the motion compensation block series and the gray level block series, the input block signal is input to the distortion calculation unit (7) nine times.

Then, the nearest block selection unit (9) compares each distortion value calculated by the distortion calculation unit (7), and selects a block signal giving the smallest distortion value from the motion compensation block signal series and the gray level block. Select from the signal series,
This is output to the subtracter (10) as the nearest block selection information and the nearest block signal as the interframe predicted value.

On the other hand, the input block signal is sent to the delay circuit (8).
In synchronization with the output timing of the nearest block signal,
It is input to a subtracter (10).

Then, the subtracter (10) outputs a differential block signal between the input block signal and the most similar block signal.

Next, the DC separation normalization calculation unit (11) calculates the average value, variance value, and DC separation normalization difference signal of the difference block signal.

Then, the validity/invalidity discriminator (12) compares the average value m and variance value σ with thresholds Th□, Th, respectively, makes the following determination, and outputs validity/invalidity information. .

For blocks that are determined to be invalid in the above determination, the human block signal and the most similar block signal are considered to be the same, and the transmission line encoder (19), which will be described later,
Only the closest block selection information is transmitted. In addition, for the blocks that have been determined to be valid, the transmission path encoder (19) transmits the most similar block selection information and the quantized encoded DC separation normalized signal.

In this case, it is understood that when both thresholds are increased, the number of invalid blocks increases and the amount of information transmitted decreases, and when they are decreased, the number of valid blocks increases and the amount of information transmitted increases.

Further, the DC separated normalized difference signal is processed by a quantization encoder (14
) is quantized by the transmission path encoder (1
9).

Then, in the case of an input block signal for which the valid/invalid information is determined to be valid, the transmission path encoder (19)
All input information is variable-length coded, and if it is invalid, only the valid/invalid information and the closest block selection information are subject to variable-length coding. Then, the transmission line encoder (
19) sends a variable-length coded signal to the transmission path frame by frame.

Then, in the threshold value control section (16), the transmission line encoder (1
Input the amount of information generated per frame in 9), calculate the threshold Th, , Th, that corresponds to the amount of information generated, and make it valid/
It is supplied to the invalidity discriminator (12). Specifically, when the amount of information generated is large, Th, , Th, is set high, and in the opposite case, Thm, Th, is controlled to be set low to keep the average amount of information generated to a predetermined value. Do something to bring it closer.

On the other hand, the quantized DC separated difference signal sequence, the average value, which is manually input to the transmission line encoder (19) as encoded data,
The deviation component, valid/invalid information, and nearest block selection information are sent to the quantization decoder (15) and the local decoder (13).
) is locally decoded by

Then, the locally decoded signal sequence passes through the in-loop filter (17) and is then written into a predetermined area of the frame memory (4) under the control of the write control section (6).

Note that the processing in the in-loop filter (17) is effective for removing isolated point noise peculiar to the block coding method, and is performed as shown in the following equation. That is, in FIG. 5, the pixel level X of one pixel in the locally decoded signal sequence is changed to the pixel level
Perform smoothing processing using .

-(1-K)

Note that, due to the characteristics of the reproduced signal that has been subjected to the above smoothing process,
Although isolated point noise has been removed, the reproduced image is blurred, and the amount of difference with the input block signal of the next frame increases, which tends to increase the amount of transmitted information of the next frame. Internal filter processing is applied only to signal sequences determined to be valid.

Therefore, as explained above, according to the image coding and transmission device that includes motion compensation, images with little movement such as in video conferences and video calls are allocated to invalid blocks when transmitting images with strong correlation between frames. This has the advantage of increasing the number of connections and enabling efficient and effective transmission.

[Problems to be Solved by the Invention] Since the conventional motion-compensated image encoding device is configured as described above, good encoding characteristics can be obtained when transmitting moving images with a small amount of motion. When transmitting a moving image with a large value, a gray level block is more likely to be selected as the closest block signal, and since the amount of information to be transmitted increases, the threshold for valid/invalid discrimination is controlled to be high, so the gray level block is selected as the closest block signal. There is a problem in that gray level blocks that have been used are often determined to be invalid, and that the boundaries of the invalid gray level blocks are visible on the reproduced image, making it difficult to obtain good visual characteristics.

This invention was made to solve the above-mentioned problems, and provides a motion compensation method that can reproduce images with good visual characteristics even for input image signals with strong motion and an increased number of invalid blocks. In order to achieve the above purpose,
The image encoding and transmitting device according to the present invention is provided with a means for monitoring the closest block selection information and valid/invalid information, and performing smoothing processing on the decoded reproduced signal, which is a gray level block as the closest block. be.

[Operation] The image encoding and transmitting device including motion compensation according to the present invention has the following features:
The gray level is selected as the closest block, and smoothing processing is also applied to blocks that are determined to be invalid.
The boundaries between gray levels become less noticeable, resulting in less visual discomfort.

Furthermore, the device of the present invention selects the motion compensation block as the most similar block and does not perform the smoothing process on the block determined to be invalid, so it is possible to suppress an increase in the amount of information generated due to the smoothing process.

[Embodiment] Hereinafter, a preferred embodiment of an image coding and transmission apparatus including motion compensation according to the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram of an image encoding and transmitting apparatus according to this embodiment. In this figure, the same parts as those in the conventional example shown in FIG. 4 are given the same reference numerals, and detailed explanation thereof will be omitted.

The features of this embodiment include an in-loop filter (17),
The in-loop filter control unit (18) performs a smoothing process on an invalid gray level block whose nearest approximate block is a gray level block.

The operation will be explained below.

The digitized image signal series of the A/D converter in the previous stage is converted into various encoded data by the transmitting side encoder (1), as in the conventional example shown in Fig. 4, and then sent to the next stage transmission line. encoder (
19), the signal is variable-length encoded and framed, and then sent to an actual transmission path.

On the other hand, the encoded data encoded by the transmitter encoder (1) is decoded by the quantization decoder (13), and then
It is input to the in-loop filter (17).

Then, the in-loop filter (17) performs 0N1 in-loop filter processing by the in-loop filter control unit (18).
It operates under the control of 0FF and constant designation.

Below, based on FIG. 2, the in-loop filter control section (1
The control operation of 8) will be explained.

First, the in-loop filter control unit (18) receives the closest block selection information and valid/invalid information.

Then, if it is a valid block, filter processing is turned on.
, the normal filter coefficient (smoothing characteristic) K1, for example -0,7, is output to the in-loop filter (17).

If the block is an invalid block, it is further determined whether the nearest block is a gray level block.

Then, for blocks that are selected as the most similar block and are determined to be invalid,
Set the loop filter processing ON instruction to the loop filter (
17). At the same time, the coefficient K for the in-loop filter processing is specified to be lower than normal (for example, K-0, 4).

On the other hand, if the block is an invalid block and the nearest block is a motion compensation block, filter processing is not performed as in the conventional case.

Therefore, the decoded signal sequence shown in FIG. 3(a) becomes as shown in FIG. 3(b) by the filtering process of the apparatus of the above embodiment.

Note that according to the device of this embodiment, the decoded signal sequence of the invalid gray level block is smoothed by the in-loop filter (17), but the filter coefficient at that time is lower than that in normal times. Since the image is less blurred, the amount of information generated in the next frame is prevented from increasing more than necessary.

[Effects of the Invention] As explained above, the image coding and transmission device including motion compensation according to the present invention performs smoothing processing on effective blocks and invalid gray level blocks using an in-loop filter during decoding and reproduction, In addition, since the filter coefficients (smoothing characteristics) during smoothing are controlled according to the valid/invalid information and the selection information of the closest block, it is possible to prevent the input image from having strong motion and many invalid blowers. This has the effect that good visual characteristics can be obtained in the reproduced image even when the image is blurred.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a preferred embodiment of an image coding and transmission device including motion compensation according to the present invention, and FIG. 2 is a flowchart of processing of the in-loop filter control unit of the device of the embodiment shown in FIG. FIG. 3 is an explanatory diagram of the operation of the in-loop filter of the embodiment device in FIG. 1, FIG. 4 is a block diagram of a conventional image encoding device including motion compensation, and FIG. 5 is an explanatory diagram of the in-loop filter. . In the figure, (1) is a transmission section, (2) is a block configuration section, (3) is a gray level generation section, (4) is a frame memory, (5) is a read control section, (6) is a write control section, (
7) is a distortion calculation unit, (8) is a delay circuit, (9) is a nearest block selection unit, (11) is a DC separation normalization unit, (12)
is a valid/invalid discriminator, (13) is a local decoder, (14
) is a quantization encoder, (15) is a quantization decoder, (1
6) is a threshold value controller, (17) is an in-loop filter, (1
8) is an in-loop filter control section, and (19) is a transmission line encoder. In addition, in the figures, the same reference numerals indicate the same or equivalent parts. Agent Patent attorney Masuo Oiwa (2 others) 2nd local decoded signal sequence (invalid blocks only in the shaded area) Figure J2) ■ ζD ■■■・knee) ■−′−] Pixels to be filtered in the loop: X In-loop filter reference pixels Two people B, C, D Figure 5 Procedural amendment (voluntary)

Claims (1)

[Claims] A preprocessing circuit that reads an image signal and blocks a predetermined number of pixels located in adjacent positions on the image; Motion compensation block generation means for generating a motion compensation block; Gray level block generation means for calculating a representative pixel level of the current input block and generating at least one gray level block consisting only of the representative pixel level; Select the closest block that is closest to the current input block from the compensation block group and the gray level block, compare the distortion values of the current input block and the closest block with a threshold, and determine if the distortion value is within the range of the threshold. When the current input block is regarded as an invalid block, only the nearest block information is encoded and transmitted, and when the distortion value is outside the threshold range, the nearest block information is regarded as a valid block, and the difference between the current input block and the nearest block is encoded and transmitted. A coding transmission means for coding and transmitting block information, and a smoothing process that captures surrounding pixel levels at a predetermined capture ratio to each pixel level of the effective block that has decoded the coded transmission information, and a decoded reproduction signal. An image encoding and transmitting apparatus including motion compensation, comprising: a smoothing means for outputting the threshold value as the encoded transmission information amount of the previous frame; and a threshold value control means for increasing the threshold value when the amount of information is large, according to the amount of encoded transmission information of the previous frame. Image encoding including motion compensation, characterized in that the smoothing means includes invalid gray level block smoothing means for performing smoothing processing on pixels whose nearest approximate block is a gray level block. Transmission device.