JPH0364189A - Moving compensation inter-frame coding device - Google Patents

Abstract

PURPOSE:To improve the quality of a reproduced picture by predicting a code quantity when a moving vector is coded to discriminate whether or not the moving vector is to be coded. CONSTITUTION:A moving vector discrimination section 50 is provided, which discriminates whether or not a moving vector detected by a moving vector detection circuit 2 is valid or invalid and predicts a coded orthogonal transformation coefficient and a code quantity of the moving vector or the code quantity of the moving vector only and also discriminates whether or not a succeeding quantization step wide is largely increased based on the predicted code quantity and the residual code quantity of a buffer 10. When a difference of the evaluation of the predicted values using the moving vector and not is between 1st and 2nd threshold levels and the succeeding quantization step width is largely increased, the moving vector is not coded. Thus, the deterioration in the quality of the reproduced picture due to an increased quantization steps is avoided and the quality of the reproduced picture is improved.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a motion compensated interframe coding device used in video conferences, video telephones, television monitoring systems, and the like.

BACKGROUND OF THE INVENTION FIG. 7 shows the configuration of a conventional motion compensated interframe coding device.

In FIG. 7, when a digitized image is input to terminal 1, motion vector detection circuit 2 detects a motion vector based on this image and the reproduced pixel values of the previous frame from frame memory 3, and detects a motion vector via switch 4. frame memory 3
is output to the encoding circuit 9.

As will be described later, the motion vector determination unit 5 determines whether this motion vector is valid or not based on the image from the terminal 1, the motion vector detected by the motion vector detection circuit 2, and the reproduced pixel value of the previous frame from the frame memory 3. It is determined whether it is invalid or not, and the switch 4 is controlled.

Further, the adder 6 subtracts the predicted value from the frame memory 3 from the image from the terminal 1 and outputs a prediction error, the orthogonal transformation circuit 7 orthogonally transforms this prediction error, and the quantization circuit 8 , the orthogonal transform coefficients transformed by the orthogonal transform circuit 7 are quantized, and the encoding circuit 9 encodes the quantized orthogonal transform coefficients and the motion vector detected by the motion vector detection circuit 2. to the line via.

Note that encoded data is read out from the line buffer 10 and sent out to the line so as to match the transmission speed of the line.
Further, the quantization step width of the quantization circuit 8 is controlled according to the amount of residual encoded data stored in the line buffer 10, so that overflow of the line buffer 10,
Underflow can be prevented.

On the other hand, the inverse orthogonal transform circuit 11 inversely quantizes the orthogonal transform coefficients quantized by the quantization circuit 8 and outputs a prediction error, and the adder 12 adds this prediction error to the predicted value from the frame memory 3. to reproduce the pixel value, and this pixel value is written into the frame memory 3.

FIG. 8 shows a detailed configuration of the motion vector determining section 5. As shown in FIG.

In FIG. 8, the evaluation amount calculation circuit 5a uses a motion vector based on the input image from the terminal 1, the motion vector detected by the motion vector detection circuit 2, and the reproduced pixel value of the previous frame from the frame memory 3. The error from the input image when the predicted value is calculated using
Find the sum of products, etc., and calculate the evaluation amount when using a motion vector.

On the other hand, the evaluation amount calculation circuit 5b similarly calculates the absolute sum of errors, the sum of squares of errors, etc. using the input image from the terminal 1 and the reproduced pixel values of the previous frame from the frame memory 3, and uses the motion vector. Calculate the evaluation amount when there is no one.

The comparison circuit 5c subtracts the evaluation amount calculated by the evaluation amount calculation circuit 5a from the evaluation amount calculated by the evaluation amount calculation circuit 5b, and when this difference is less than a predetermined value, the motion vector detection circuit 2 detects the evaluation amount. Invalidates the motion vector.

Therefore, when the motion vector detected by the motion vector detection circuit 2 is invalidated, the encoding circuit 9 encodes only the quantized orthogonal transform coefficients, thereby suppressing the amount of generated code. I can do it.

Problems to be Solved by the Invention However, in the conventional motion-compensated interframe coding device described above, the motion vector determining unit 5 determines whether a motion vector is valid or invalid based only on an evaluation amount such as a prediction error. If the vector is determined to be valid, the amount of code encoded by the encoding circuit 9 increases, and therefore the amount of residual encoded data accumulated in the line buffer 710 increases, and the quantization step width of the quantization circuit 8 increases. There is a problem in that the size of the image increases and the quality of the reproduced image deteriorates.

SUMMARY OF THE INVENTION In view of the above-mentioned conventional problems, it is an object of the present invention to provide a motion compensated interframe coding device that can improve reproduced images.

Means for Solving the Problems In order to achieve the above object, the present invention predicts the code amount of orthogonal transform coefficients and motion vectors to be encoded, or the code amount of only the motion vector, and also combines the predicted code amount and , determine whether the next quantization step width will increase significantly depending on the amount of residual code in the buffer, and calculate the evaluation amount of the predicted value when using a motion vector and the evaluation amount of the predicted value when using no motion vector. When the difference is between the first and second thresholds and the next quantization step width increases significantly, the motion vector is not encoded.

Effect of the Invention With the above configuration, the present invention predicts the code amount when a motion vector is encoded and determines whether or not to encode a motion vector, so that the quantization step width increases and the reproduced image quality deteriorates. Therefore, the reproduced image can be improved.

EXAMPLES Hereinafter, the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of a motion compensated interframe coding device according to the present invention, FIG. 2 is a block diagram showing a detailed configuration of the motion vector determining section in FIG. 1, and FIG. is the second
FIG. 4 is an explanatory diagram of the operation of the generated code amount prediction circuit shown in FIG.
8 is an explanatory diagram of the operation of the motion vector determination circuit shown in FIG. 7, and the same reference numerals are given to the same components as those shown in FIG. 7. FIG.

In FIG. 1, 1 is an input terminal for a digitized image, and 2 is an input terminal for an image from terminal 1 and a frame memory 3.
This is a motion vector detection circuit that detects a motion vector based on the reproduced pixel value of the previous frame from , and outputs it to the frame memory 3 and the encoding circuit 9 via the switch 4 .

50, as described later, represents an image from the terminal 1, a motion vector detected by the motion vector detection circuit 2, a reproduced pixel value of the previous frame from the frame memory 3, and a quantization step from the quantization circuit 8a. This is a motion vector determination unit that determines whether the motion vector detected by the motion vector detection circuit 2 is valid or invalid based on the congestion and the amount of residual encoding stored in the line buffer 10.

Note that when the motion vector determination unit 50 determines that the motion vector is valid, the encoding circuit 9 encodes the motion vector together with orthogonal transformation coefficients, and when it is determined that the motion vector is invalid, the encoding circuit 9 encodes the motion vector together with orthogonal transformation coefficients. Encode only the coefficients.

6 is an adder that subtracts the predicted value from frame memory 3 from the image from terminal 1 and outputs a prediction error; 7 is an orthogonal transform circuit that orthogonally transforms this prediction error by discrete cosine transform, etc.; 8a is , is a quantization circuit that quantizes the orthogonal transform coefficients transformed by the orthogonal transform circuit 7, and the encoding circuit 9 quantizes the quantized orthogonal transform coefficients and the motion vector detected by the motion vector detection circuit 2. Both or only the former are encoded and sent to the line via the line buffer 10.

Encoded data is read out from the line buffer 10 so as to match the transmission speed of the line and sent out to the line, and the quantization circuit 8a reads out the encoded data in accordance with the amount of residual encoded data stored in the line buffer 10. The quantization step width of is controlled, thus preventing an overflow of the line buffer 10,
Underflow can be prevented. Further, this residual encoded data amount is supplied to the motion vector determination circuit 50 as described above.

11 inversely quantizes the orthogonal transform coefficients quantized by the quantization circuit 8a. 11 is an inverse orthogonal transform circuit that outputs the difference, and 12 is an adder that adds this prediction error and the predicted value from the frame memory 3 to reproduce a pixel value, and writes this pixel value to the frame memory 3. .

Here, the prediction error dequantized by the inverse orthogonal transform circuit 11 includes the quantization error of the quantization circuit 8a, and is therefore different from the prediction error output by the adder 6.

In FIG. 2, 51 is based on the input image from the terminal 1, the motion vector detected by the motion vector detection circuit 2, and the reproduced pixel value of the previous frame from the frame memory 3.
Calculate the error from the input image when a predicted value is calculated using a motion vector, that is, the absolute sum of errors, the sum of squared errors, etc.
An evaluation amount calculation circuit 52 that calculates an evaluation amount when a motion vector is used similarly calculates the absolute sum of errors and the error amount using the input image from the terminal l and the reproduced pixel value of the previous frame from the frame memory 3. This is an evaluation amount calculation circuit that calculates the sum of squares, etc., and calculates the evaluation amount when no motion vector is used.

53 is the quantization step width q1, as shown in FIG.
For each q2...qn, the generated code amount corresponding to the evaluation amount calculated by the evaluation amount calculation circuit 51 is determined and set in advance experimentally or by calculation, and the motion vector detected by the motion vector detection circuit 2. A generated code amount prediction circuit predicts the amount of code generated after encoding by the encoding circuit 9, as described later, based on the evaluation amount calculated by the evaluation amount calculation circuit 51 and the quantization step width of the quantization circuit 8a. It is.

A quantization step width 54 detects how the quantization step width of the quantization circuit 8a is updated based on the code amount predicted by the generated code amount prediction circuit 53 and the residual code amount of the line buffer 710. The update detection circuit 55 subtracts the evaluation amount calculated by the evaluation amount calculation circuit 52 from the evaluation amount calculated by the evaluation amount calculation circuit 51, and uses this difference and the quantization step width update detection circuit 54 as described later. Due to the magnitude of the quantization step width update detected by
This is a motion vector determination circuit that determines whether the motion vector detected by the motion vector detection circuit 2 is valid or invalid.

Next, the operation of the above embodiment, particularly the operation of the generated code amount prediction circuit 53 and the motion vector determination circuit 55 of the motion vector determination section 50, will be described with reference to FIGS. 3 and 4.

First, in FIG. 1, when the orthogonal transform coefficients quantized by the quantization circuit 8a and the motion vectors detected by the motion vector detection circuit 2 are encoded by the encoding circuit 9, the amount of code increases. When the amount of residual codes in the line buffer 710 increases, the quantization step width of the quantization circuit 8a increases, and quantization noise increases.

Therefore, the generated code amount prediction circuit 53 of the motion vector determination circuit 500 uses the evaluation amount calculated by the evaluation amount calculation circuit 51, the quantization step width of the quantization circuit 8a, and the quantization step width set in advance as shown in FIG. The amount of code after encoding the orthogonal transform coefficients and the amount of encoding for encoding the motion vector detected by the motion vector detection circuit 2 are predicted from the graph.

Note that since the evaluation amount calculated by the evaluation amount calculation circuit 51 corresponds to the magnitude of the prediction error, the code amount can be predicted based on the quantization step width.

Next, the quantization step width update detection circuit 54 determines how the quantization step width of the quantization circuit 8a is updated based on the code amount predicted by the generated code amount prediction circuit 53 and the remaining code amount of the line buffer 10. When the motion vector determination circuit 55 detects the difference, the motion vector determination circuit 55 subtracts the evaluation amount calculated by the evaluation amount calculation circuit 52 from the evaluation amount calculated by the evaluation amount calculation circuit 51.

As shown in FIG. 4, the motion vector determination circuit 55 validates the motion vector when the difference is extremely large, invalidates it when the difference is extremely small, and determines the motion vector when the difference is within a predetermined width. If the amount of variation in the quantization step width is large even if the quantization step width becomes large, the motion vector is determined to be invalid.

In other words, when the amount of variation in the quantization step width is large, quantization noise increases, but in this case, the signal vector is invalidated and no encoding is performed, so the amount of residual code in the line buffer 710 is reduced and the quantization The step width of the circuit 8a becomes smaller, and therefore the reproduced image quality can be improved.

Next, a second embodiment of the present invention will be described with reference to FIGS. 5 and 6.

In this embodiment, the motion vector code amount calculation circuit 53a is
The motion vector detection circuit 2 calculates the amount of code necessary to encode and transmit the detected motion vector, and the quantization step width variation detection circuit 54 calculates the amount of code necessary for encoding and transmitting the motion vector detected by the motion vector detection circuit 2. Using the code amount, it is detected whether the quantization step width is updated when the motion vector is encoded.

As shown in FIG. 6, the performance vector determination circuit 55a calculates the evaluation amount calculated by the evaluation amount calculation circuit 51,
The ratio of the evaluation amounts calculated by the evaluation amount calculation circuit 52 is “1”.
”, it is determined to be invalid, and when it is close to “O”, it is determined to be valid, and if the ratio of the evaluation amounts is within a predetermined radius α, as the next quantization step width becomes larger, shall be invalidated.

By setting this convergence a to a value when the code amount of the quantized orthogonal transform coefficient is smaller than the code amount of the motion vector, the motion vector code amount and the residual code amount of the line buffer 7 can be used. , it is possible to detect whether the quantization step width is updated based on the amount of motion vector code.

Therefore, in this embodiment as well, the reproduced image quality can be improved, but the apparatus can be simplified compared to the previous embodiment.

As described in detail, the present invention predicts the code amount of orthogonal transform coefficients and motion vectors to be encoded, or the code amount of only the motion vector, and uses the predicted code amount and the residual code of the buffer. It is determined whether the next quantization step width increases significantly depending on the amount, and the difference between the evaluation amount of the predicted value when using a motion vector and the evaluation amount of the predicted value when using no motion vector is determined as the first, Since the motion vector is not encoded when the next quantization step width increases significantly when it is between the second threshold, the playback quality does not deteriorate due to the increase in the quantization step width. , Therefore, the reproduced image can be improved.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of a motion compensated interframe coding device according to the present invention, FIG. 2 is a block diagram showing a detailed configuration of the motion vector determining section in FIG. 1, and FIG. is an explanatory diagram of the operation of the generated code amount prediction circuit of FIG. 2, FIG. 4 is an explanatory diagram of the operation of the motion vector determination circuit of FIG. 2, and FIG. 5 is an explanatory diagram of the motion compensation interframe coding apparatus according to the present invention. FIG. 6 is an explanatory diagram of the operation of the motion-compensated interframe coding device of FIG. 5, and FIG. 7 shows a conventional motion-compensated interframe coding device. Block Diagram,
FIG. 8 is a block diagram showing the detailed configuration of the motion vector determination circuit of FIG. 7. 2... Motion vector & drink circuit, 3... Frame memory, 6, 12... Adder, 7... Orthogonal transformation circuit, 8a
Quantization circuit, 9 Encoding circuit, 10 Transmission buffer, 11 Inverse orthogonal transform circuit, 50 Motion vector determination unit, 51.52 Evaluation amount calculation circuit ,5
3... Generated code amount prediction circuit, 53a... Motion vector code amount calculation circuit, 54... Quantization step radius update detection circuit, 54a... Quantization step width variation detection circuit, 5
5゜552--Movement vector determination circuit.

Claims (2)

[Claims] (1) means for calculating a prediction error based on a predicted value of a video signal; means for orthogonally transforming the prediction error to calculate a coefficient; means for quantizing the orthogonal transformation coefficient; and a motion vector of the video signal. a means for detecting the quantized orthogonal transform coefficients and a motion vector; an encoding means for encoding the quantized orthogonal transform coefficients and the motion vector; and a means for temporarily accumulating and outputting the code encoded by the encoding means. a buffer that controls the quantization step width of the quantization means according to the amount of residual code; a first calculation means that calculates an evaluation amount of a predicted value when the motion vector is used; and a first calculation means that does not use the motion vector. a second calculation means for calculating an evaluation amount of a predicted value for a case; a motion vector detected by the motion vector detection means; an evaluation amount of the first calculation means; and a quantization step width of the quantization means. means for predicting the code amount of orthogonal transform coefficients and motion vectors to be encoded by the encoding means; and the next quantization step width is greatly increased by the predicted code amount and the residual code amount of the buffer. a means for determining whether or not to perform a motion vector; and controlling the encoding means to encode the motion vector when the difference between the evaluation amounts of the first and second calculation means is larger than a first threshold; If the difference is smaller than a second threshold value which is smaller than the first threshold value, the encoding means is controlled not to encode the motion vector, and this difference is smaller than the first threshold value and the second threshold value.
and means for controlling the encoding means so as not to encode the motion vector when the next quantization step width increases significantly when the width is between the threshold values. (2) Instead of predicting the code amount of the orthogonal transform coefficient and motion vector encoded by the encoding means, the code amount when only the motion vector is encoded is predicted. ) The motion compensated interframe coding device described in (a).