JPH0385882A - Image signal coding device - Google Patents

Abstract

PURPOSE:To improve coding efficiency by measuring the variation of each picture element signal, and when a period in the state that the absolute value of the variation is less than a previously determined range is longer than a previously determined value, stopping the coding of a forced intra-frame signal. CONSTITUTION:A signal quantized by a quantizing circuit 5 is converted into an absolute value in each picture element by an absolute-valuing circuit 11, the absolute value is compared with a previously determined threshold by a compartor 12, and when the absolute value exceeds the threshold, the generation of variation is decided and a counter circuit 13 is reset. A comparator 15 compares the output of the circuit 13 with a previously determined threshold, and when the output exceeds the threshold, sends a forced intra-frame signal stop signal to change the connection of a switch and stop a forced intra-frame signal. Consequently, the coding efficiency of the coding device can be improved.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to high-efficiency coding of television signals using interframe prediction, and in particular, to reduce the propagation of transmission errors in the time axis direction. This paper relates to a transmission error mitigation method that inserts coding based on certain rules.

[Conventional technology]

The "interframe prediction method" is known as one of the methods for highly efficient encoding of television signals. This method calculates the difference between the image frame that was transmitted immediately before (hereinafter referred to simply as a frame) and the frame that is currently being transmitted, and then applies quantization and variable length encoding to this to perform high-efficiency encoding. .

Since there is often little movement between frames in a television signal, the power of the difference signal between frames is significantly smaller when compared with the original television signal. Especially in static areas such as the background of the screen, the difference between frames is sufficiently small.

Normally, this difference is not transmitted, and only the fact that there is no difference is notified (referred to as an invalid pixel). Since a plurality of invalid pixels usually occur in quick succession, the number of consecutive invalid pixels, etc. is encoded with high efficiency. As a result, image signals can be compressed even more efficiently.

As a result, if the intra-frame signal that is currently being transmitted is encoded as is, good image quality cannot be obtained unless a coded signal of about 3 bits is allocated per pixel, whereas inter-frame prediction is It is known that when used, it can be compressed to about 1 bit.

By the way, in interframe prediction, the signal of the next frame is predicted using the previous frame that has already been transmitted, so if the image signal reproduced in the previous frame is incorrect due to a -degree transmission error, there will be no transmission error in the next frame. Errors occur in the reproduced signal. This is called propagation of transmission errors in the time axis direction. In principle, there is a problem that this error propagates infinitely.

The following methods are known as methods for reducing this error propagation.

(1) Periodic refresh - Forcibly applies intraframe signal encoding at a fixed period to stop propagation of transmission errors in the time axis direction.

(2) Dayman Refresh Uni Calculate the parity of the signal obtained by restoring the image signal of one frame on the transmitting side and the receiving side, and notify the receiving side of the result from the transmitting signal. On the receiving side, the parity notified from the transmitting side is compared with the parity of the frame reproduced on the receiving side to check whether a transmission error has occurred.

When it is determined that a transmission error has occurred, the receiving side requests the transmitting side to transmit again.

[Problem to be solved by the invention]

In the periodic refresh described in (1) above, the intra-frame signal is encoded at a constant period, regardless of whether the image signal was transmitted due to inter-frame pre-holi during this period or whether it was never transmitted as an invalid pixel. be converted into For this reason, a forced intra-frame signal having a codeword length longer than necessary is applied, resulting in a problem of reduced coding efficiency.

If the period of this periodic refresh is lengthened, the encoding efficiency can be improved, but if a -degree transmission error occurs, a problem arises in that it takes a long time to restore the correct image.

In (2) dayman refresh, intraframe coding is applied only when necessary to prevent the propagation of transmission errors, so it is notified whether the coding efficiency is high or not, and then the transmitting side uses the intraframe signal to Because of switching, there is a problem that it takes a long time to restore the correct image.

Furthermore, when transmitting from one transmitter to a plurality of receivers, the transmitting side needs to apply an intra-frame signal even when a transmission error occurs in one receiver. Therefore, it is an object of the present invention to provide a method for reducing the number of refreshes and reducing the propagation of transmission errors in the time axis direction in order to improve image quality.

[Means to solve the problem]

In order to achieve the above object, the present invention has the following features: (1) On the transmitting side, transmission errors occur in units of one screen, or units in which a screen is divided into multiple areas, or units in which a transmission signal is divided into packets. means for adding an additional signal to allow the receiving side to determine whether or not a transmission error has occurred; means for forcibly sending an intra-frame signal at a predetermined period to suppress propagation of transmission errors in the time axis direction; means for measuring the time during which the absolute value of the amount of variation is below a predetermined range; and means for stopping the forced in-frame signal as described in (ii) when this time is longer than the predetermined value. , (3) On the receiving side, determine whether a transmission error has occurred in each screen or divided area,
The present invention is characterized by comprising means for replacing a portion where a transmission error has been detected with a signal of a screen transmitted immediately before.

[Effect] Assuming that there is a frame in which it is considered that no transmission error has occurred, (1) In all subsequent frames, if a pixel is determined to be an invalid pixel because the difference between frames is sufficiently small, Even if a transmission error occurs, the decoded image is virtually unaffected by the transmission error. Therefore, refreshing using an intra-frame signal is not necessary for this pixel.

(2) Also, even if inter-frame prediction is performed between them,
When the interframe difference between the input signal and the predicted signal is small, refreshing using the intraframe signal can be omitted in the following cases.

This is because even if the signal is replaced by the previous screen signal on the receiving side due to a transmission error, the difference between the transmitted frames is also small.
The difference from the image signal (true signal) obtained by correctly decoding the image signal is small.

(3) On the other hand, when the difference between frames is large, the signal is replaced with the signal of the previous screen due to a transmission error.

The difference from the true image signal is large, resulting in significant image quality deterioration. Therefore, when the signal is displayed for a certain period of time or more, it is refreshed by the intra-frame signal.

〔Example〕

An embodiment of the present invention will be described below with reference to FIG.

FIG. 1 exemplifies a case in which, in the present invention, when the signal value of a pixel in a frame changes from frame to frame, intra-frame No. 1 encoding is forcibly performed.

To explain the figure, (]) The horizontal axis of the figure represents the frame number, and the vertical axis represents the signal value of the pixel.

(2) The white circles ○ represent the normal inter-frame secondary side, and the black circles and
indicates the case where the signal is forced to be an intra-frame signal.

(3) A solid line shows a case where a signal that was correctly received without a transmission error is decoded, and a broken line shows a case where a transmission error occurs and the signal is replaced with the signal of the previous frame on the receiving side.

If a transmission error occurs at time t1 when the signal value is rapidly changing, the influence of the transmission error is suppressed by the forced intra-frame signal. The same effect as the conventional periodic refresh illustrated in FIG. 2 can be obtained. Furthermore, since an intra-frame signal is used in this period in the same manner as in periodic refresh, the encoding efficiency is also the same.

If a transmission error occurs at time t2 when the confidence value is changing slowly, unlike conventional periodic refresh, the present invention does not apply forced intra-frame signal encoding. Therefore, the propagation of the shadow of transmission errors is not suppressed.

However, as is clear from the figure, the change in the reliability value is gradual, so the difference between the "true image signal" and the "image signal replaced with the signal of the previous frame" is sufficiently small. Image quality deterioration is difficult to detect visually. Furthermore, since no forced intraframe signal is applied during this period, encoding efficiency can be improved.

Next, the operation of the encoding circuit for carrying out the present invention will be explained using the block diagram shown in FIG.

In the figure, the area surrounded by dotted lines is related to the present invention, and the other parts are a block diagram of a conventional interframe encoding device that performs periodic refreshing. First, the conventional part will be briefly explained.

The image signal read from TV camera 1 is analog 0g/
The digital converter 2 digitizes the signal to generate a pixel-by-pixel signal. The subtraction circuit 3 obtains an interframe prediction error signal, which is a difference between the pixel signal and the signal of the previous frame stored in the frame memory 4.

This interframe prediction error signal is quantized by a quantization circuit 5 and variable length encoded by a variable length encoding circuit 6. If the quantization results in a non-zero value, a variable length code word corresponding to that value is assigned. Furthermore, if the quantization result is zero, the number of consecutive zeros is calculated, and that number is variable-length coded.

” The signal quantized in Section 2 is dequantized by the dequantization circuit 7 and converted into the original interframe prediction error signal,
The adding circuit 8 adds the signal of the previous frame read from the frame memory and stores the result in the frame memory 4. The signal stored in the frame memory 4 is used to obtain the interframe prediction error of the pixel signal of the next frame.

The forced intra-frame signal switching circuit 9 sends out a control signal every predetermined frame period, switches the switch 10, and replaces the output of the frame memory 4 with a fixed value.

As a result, the input image signal itself, ie, the intra-frame signal, is encoded.

The control signal is also sent to the variable length encoding circuit 6, and notifies the receiving side that the forced intra-frame signal has been encoded. The above switching can be done on a screen-by-screen basis, or
It may be done pixel by pixel.

Next, the details within the dotted line related to the present invention will be described in detail. The signal quantized by the quantization circuit 5 is converted to an absolute value pixel by pixel by the compression value conversion circuit 11, and compared with a predetermined threshold value by the comparison circuit 12. If the absolute value is equal to or greater than the threshold value, the signal is changed. As a result, the step circuit 13 is reset.

The progress circuit 13 reads “
The number of frames in which the absolute value was less than the threshold in the pixel one frame before is added by 1. However, when it receives a reset signal from the comparison circuit 12, it is reset to O. The result is stored in the frame memory 14.

The comparison circuit 15 compares the output of the step circuit 13 with a predetermined threshold, and if it is equal to or greater than the threshold, sends out a forced intra-frame signal stop signal, and switches the switch 16 to stop the forced intra-frame signal. .

It is clear that the following modifications are also within the scope of the present invention.

(1) In the above embodiment, a case has been described in which interframe predictive coding is performed for each pixel, but the present invention is not limited to this. Block coding: ``In other words, a method in which multiple prediction error signals are combined into a block, which is subjected to orthogonal transformation such as Fourier transformation or discrete cosine transformation, and then quantized.''
Alternatively, it may be applied to "vector quantization in which a plurality of interframe prediction error signals are collectively quantized."

In this case, it is easy to realize this by determining whether there is a significant difference in each block and, if the absolute value of the difference is small, determining that there is no change and stopping the forced intra-frame signal encoding.

(2) When applying periodic refresh to each screen, each pixel, or each block mentioned above,
The present invention is applicable.

(3) Even for regions determined to have no fluctuations, periodic refresh may be performed at long intervals as long as the encoding efficiency is hardly reduced.

This makes it possible to suppress the propagation of transmission errors even in areas where the influence is small.

(4) In a predictive coding method that uses interframe/intraframe signals together, if there is no large difference between the error power due to interframe prediction and the power of the intraframe signal itself, intraframe signals may be selected. . By employing intraframe signals, propagation of the effects of transmission errors can also be suppressed.

The phase within the forced frame for subsequent periodic refresh may be initialized based on the point in time when the intraframe signal is adopted.

(5) If the frame difference in the frame with frame number i is di, and the probability that a transmission error will occur in each frame is P, then the expected value of the influence of a transmission error can be expressed as follows. Here, i=O is the number of the frame encoded as an intra-frame signal, and 10 is the intra-frame signal currently being forced to be decoded.

〔Effect of the invention〕

According to the above invention, it is possible to reduce the number of forced borrowed codes within one frame while maintaining the same effect of suppressing the propagation of transmission errors as in the prior art, thereby improving coding efficiency. Therefore, great effects can be obtained especially when applied to "packet transmission, etc." in which transmission errors occur frequently and it is easy to accurately notify the receiving side of the presence or absence of transmission errors.

[Brief explanation of drawings]

FIG. 1 is a diagram illustrating fluctuations in image signal values in the time axis direction and a decoded signal when a transmission error occurs, for explaining the present invention. FIG. 2 shows an example of an image signal when conventional periodic refresh is adopted and a decoded signal when a transmission error occurs. FIG. 3 is a block diagram of an interframe predictive coding device implementing the present invention. Explanation of symbols 1...TV camera, 2...Analog/digital converter, 3...Subtraction circuit, 4, 14...Frame memory, 5...Quantization circuit, 6...Variable length encoding circuit,
7... Inverse quantization circuit, 8... Addition circuit, 9... Forced intra-frame signal switching circuit, 10.16... Switch, 11. Absolute value conversion circuit, 12, 1.5... Comparison circuit, ]-3...Step circuit. ↑ Eye number t // Otsu 2 sr: I glaze number

Claims (1)

[Claims] 1. On the transmitting side, means for accumulating at least one screen of image signals such as television signals, predicting the next input image signal using the image signal, and calculating the interframe prediction error signal, which is the difference, to a certain threshold value. If the difference is smaller than a threshold, the receiving side is notified that the difference is small. In other cases, the inter-frame prediction error signal or the input intra-frame signal itself is converted into a high-efficiency compression code. In order to make the receiving side determine that a transmission error has occurred in units of one screen, in units of dividing the screen into multiple areas, or in units of dividing the transmission signal into packets, etc. A means for transmitting the additional signal, a means for forcibly switching to intra-frame signal encoding at regular intervals to prevent the influence of transmission errors from propagating infinitely in the time axis direction, and a means for transmitting the above-mentioned signal on the receiving side. means for receiving and decoding it into the original image signal; means for determining the presence or absence of a transmission error for each unit using the additional signal; means for replacing the image signal with a signal; means for measuring the amount of variation in the image signal for each pixel signal; and means for measuring the time during which the absolute value of the amount of variation is below a predetermined range; An image signal encoding device characterized by comprising means for stopping the forcible encoding of the intra-frame signal when this time is longer than a predetermined value.