JPH05130416A - Orthogonal transformation coder - Google Patents

Abstract

PURPOSE:To obtain the orthogonal transformation coder in which data quantity control over blocks is facilitated and the increase in the circuit scale is very less, with respect to the orthogonal transformation coder used for applying high efficiency coding to a video signal. CONSTITUTION:The coder is provided with a coefficient and rearrangement device 3 used to encode a transformation coefficient subject to orthogonal transformation, a classifying device 60 classifying the orthogonal transformation coefficient into a prescribed class in parallel with the rearrangement device 3, a maximum class detector 71 detecting the class to the maximum transformation coefficient and a latch 72 latching the maximum class and the latch device 72 fixes an input signal of the classifying device 60 to a prescribed level.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an orthogonal transform coding apparatus used to increase the compression rate when a video signal is coded with high efficiency.

[0002]

2. Description of the Related Art A band compression technique (or a high-efficiency code) for reducing the data amount of an original video signal for long-time recording in a recording / reproducing device (VTR, video disc, etc.) for recording / reproducing a video signal. Technology) is used.
Orthogonal transform coding is one of the band compression techniques. Orthogonal transform coding is a technique in which a video signal is divided into blocks and frequency decomposition is performed in block units to reduce the amount of data to be transmitted (recorded) for higher frequency components. It utilizes the visual characteristic that is difficult to detect.

FIG. 3 shows a block diagram of a conventional orthogonal transform coding apparatus for performing the above orthogonal transform coding, and its operation will be described.

In FIG. 3, 1 is an input terminal for inputting a blocked signal of a video signal, 2 is an orthogonal transformer for orthogonally transforming the blocked signal, and 3 is an orthogonal transform coefficient obtained from the orthogonal transformer 2. Therefore, a rearranger for rearranging the sequence of transform coefficients, 4 is an encoder for encoding the orthogonal transform coefficient of the output of the rearranger 3, and 5 is an output terminal of the orthogonal transform encoder for encoding data. Output. Further, the encoder 4 includes a quantizer 41 for quantizing the orthogonal transform coefficient with a certain step width and a quantizer selector 42 for selecting a quantizer having a step width suitable for containing a desired amount of data.
And an encoder 43 for encoding the quantized data output from the quantizer 41.

The encoder 43 uses variable length coding in which a code word having a higher occurrence frequency is assigned a shorter code word (smaller data amount) with respect to the occurrence frequency of the coded code word.
Therefore, the quantization selector 42 selects the quantizer 41 having the optimum step width by calculating the data amount of the result of the variable length coding.

FIG. 4 shows the arrangement of transform coefficients in block units in the orthogonal transformer 2 output A and the rearranger 3 output B for explaining the operation of the rearranger 3. The coefficient arrangement shown in FIG. 4 is obtained by the orthogonal transform coding apparatus of the conventional example, which adopts a method of two-dimensional orthogonal transform in the horizontal direction and the vertical direction.
The block has a block size of 4 horizontal pixels and 4 vertical pixels. Therefore, in the signal A in FIG. 4, one block is composed of 16 coefficients of 4 coefficients in the horizontal direction and 4 coefficients in the vertical direction. In the block of FIG. 4, the frequency component represented by each coefficient corresponds to the lower horizontal band in the leftward direction and to the lower vertical band in the upper direction. In the configuration of the orthogonal transformer 2 that performs two-dimensional orthogonal transformation, since the orthogonal transformation in one direction in the horizontal direction and the vertical direction in the other direction is performed, the orthogonal transformation in the other direction is performed.

However, in order to perform the coding for the two-dimensional orthogonal transformation, the coefficient of the signal B and the two-dimensional frequency low range called zigzag scan as shown (horizontal and vertical low ranges in the upper left of the figure). It is suitable to line up in the high range (lower right of the figure). In other words, the low-frequency component including the DC component has a greater effect on the visual sense, and the lower-frequency component is more important. Therefore, in the encoder 4, the step width may be made finer from the beginning of the coefficient sequence until the data amount reaches the desired data amount, starting from the state of quantization with the coarsest step width.

[0008]

However, the above-mentioned structure has the following problems. In terms of the data amount by encoding, a block having a large data amount includes a block including a large amplitude frequency component, and a block having a small data amount includes a block including only a small amplitude frequency component. If similar step widths are assigned to these two types of blocks to reduce the amount of data,
Most of the coefficients of the latter block containing only small amplitude frequency components are deleted (coefficients are assigned to 0). This is a block of a flat screen, but important information is lost and block noise is generated (for example, noticeable in the background vegetation and the sky). On the other hand, the former block including a large-amplitude frequency component is not reduced by the step width because of the amplitude of the frequency component, and thus the influence of the quantization error is visually smaller than the latter block.

However, since the optimum quantization performed in the encoder 4 of the conventional orthogonal transform encoder is performed for each block, it is difficult to control the selection of the quantizer between blocks (data amount control). Then, the quantization selection operation is performed in the same manner as the block including the large amplitude frequency component in the block including only the small amplitude frequency component, and the image quality of the flat portion such as the background is deteriorated. ..

In view of the above problems of the prior art, it is an object of the present invention to provide an orthogonal transform coding apparatus that facilitates control of the data amount between blocks, increases the circuit size, and consumes very little power. And

[0011]

According to the present invention, an orthogonal transforming means for orthogonally transforming a block-shaped signal, a coefficient for encoding a transform coefficient orthogonally transformed by the orthogonal transforming means and a rearrangement for each block. Sorting means, classifying means for classifying the transform coefficient into a predetermined class according to a plurality of thresholds, and maximum class for detecting the maximum class having the maximum value of transform coefficients among the classes obtained by the classifying means for each block. An orthogonal transform coding device characterized in that the class detecting means, the classifying means after the maximum class detection, and the maximum class detecting means are stopped.

[0012]

With this configuration, the result of classifying the transform coefficient corresponding to the maximum amplitude of the frequency components contained in each block is obtained, so that the amplitude of the frequency component for each block can be easily known at the time of the subsequent encoding. Even with a small amplitude, the amount of data can be controlled without reducing important frequency components. Furthermore, since the class having the maximum value of the conversion coefficient is held and then the operation is stopped, the power consumption is small.

[0013]

1 is a block diagram of an orthogonal transform coding apparatus according to an embodiment of the present invention, and FIG. 2 is a change diagram of transform coefficients and classes.

In FIG. 1, reference numeral 1 denotes an input terminal of this embodiment, which has a block size of horizontal 4 pixels and vertical 4 pixels per block as in the conventional example. Reference numeral 2 is an orthogonal transformer, which performs two-dimensional orthogonal transformation of the input block of 4 × 4 pixels, and 3 is the orthogonal transformer 2 for performing two-dimensional frequency quantization and coding in the encoder of 4. It is a rearranger that rearranges the array of output conversion coefficients so as to be arrayed from a low band to a high band at a two-dimensional frequency. An output terminal 5 outputs the orthogonal transform coefficient encoded by the encoder 4.
A detector 70 having a classifier 60, a maximum class detector 71, and a holder 72 is newly provided.

In FIG. 2, the horizontal axis represents the order of coefficients, which corresponds to the order A of conversion coefficients in the conventional example of FIG.
The vertical axis on the left is the magnitude of the amplitude, which is the absolute value of the conversion coefficient. The vertical axis on the right is three thresholds of P1, P2 and P3 and 4
It is a class of kind. The four types of classes are 0, 1, 2, and 3 from the smallest absolute value of the conversion coefficient. The first conversion coefficient of one block is a DC component, and the classification is set to 0.

The classifier 60 uses the orthogonal transform coefficient A shown in FIG.
Is input, the conversion coefficients having smaller absolute values are classified into four classes of 0, 1, 2, and 3 according to the magnitude of the absolute value of each orthogonal conversion coefficient. The result is shown in C'of FIG.

Next, the maximum class detector 71 detects the maximum class corresponding to the maximum conversion coefficient based on the class C '(D in FIG. 2) and holds the maximum class for one block in the holder 72. 4 to the quantization selector 45 in the encoder.
Here, the holder 72 holds the maximum class and then fixes the input signal (conversion coefficient) classified into classes to a predetermined level "HI" or "LOW" regardless of the input signal.
When fixed to "HI", the retainer 72 outputs "LOW" until the maximum class is detected, and "H" is detected at the time when the maximum class is detected.
It is sufficient to output "I" and take "OR" with the input signal of the classifier 60. Further, when fixing to "LOW", the retainer 72 detects "HI" until the maximum class is detected. At this point, “LOW” is output and “AND” is taken with the input signal of the classifier 60. The change in the output of the classifier 60 when the input signal is fixed as described above is shown in FIG.

According to the present embodiment as described above, the quantization selection is performed based on the class having the maximum value of the conversion coefficient (excluding the DC component) for each block obtained by the maximum class detector 71 and the holder 72. The device 45 can prevent the deterioration of the image quality in the flat portion by lowering the ratio of deleting the conversion coefficient of the block including only the small amplitude frequency component as compared with the block including the large amplitude frequency component.

If a large number of classes for large transform coefficients are allocated as in the present embodiment, it is only necessary to find the maximum value of the number of allocations. In addition, the range of the input class is 0 to 3 in the present embodiment, which is 2 bits, and the range of the orthogonal transform coefficient is generally required to be 8 bits or more, so that the maximum class candidate is detected. The comparator and the register for storing the maximum class candidate itself are small, which is very effective.

Further, by fixing the input signal for classifying to a predetermined level when the maximum class is held, the output of the classifier 60 is also fixed, so that the classifier 60 and the maximum class detector 71 are fixed. There is no change in input level and current consumption is reduced. Also, the classifier 60
The same effect can be obtained by fixing the input signal of the maximum class detector 71 as well as the input signal of. The same effect can be obtained without stopping the operation after holding.

Here, what excludes the DC component from the classification target is the offset of the entire block of the AC component which is composed of other frequency components, and the DC component is the amplitude value of the frequency component at present. Differently, it is not related to image quality deterioration.

In the embodiment of the present invention, the block signal to be orthogonally transformed has a block size of 4 × 4 pixels, but it may have a block size of 8 × 8 pixels or 16 × 16 pixels. Furthermore, instead of horizontal and vertical two-dimensional orthogonal transformation, three-dimensional orthogonal transformation may be used. As described above, as the block size or the number of dimensions increases and the number of transform coefficients forming a block increases, the maximum class of transform coefficients is detected in parallel with the rearranger 3 as in the present embodiment. It is more effective to carry out by executing the above because a delay device or the like is not newly required for the detection. Further, although the operation is stopped by fixing the input level regardless of the input signal, it may be fixed according to the input. Further, even if the power supply and the clock are stopped, the same effect can be obtained.

[0023]

As described above, according to the present invention,
Since the result of classifying the maximum value of the transform coefficient corresponding to the maximum amplitude of the frequency components included in each block is obtained, the amplitude of the frequency component of each block can be easily known at the time of encoding in the subsequent stage, and even with a small amplitude. Since the amount of data can be controlled without deleting important frequency components, orthogonal transform coding can be performed without image quality deterioration, which is very effective. Furthermore, since the class of the frequency component is obtained in parallel with the rearranger, no new memory is needed for class detection, which is advantageous in terms of circuit scale, and after maximum class detection. Since the operation is stopped, the power consumption is small and the practical effect is large.

[Brief description of drawings]

FIG. 1 is a block diagram of an orthogonal transform coding device according to an embodiment of the present invention.

FIG. 2 is a change diagram of the absolute value of a conversion coefficient for explaining the first upper example.

FIG. 3 is a block diagram of a conventional orthogonal transform encoding device.

FIG. 4 is an arrangement diagram of coefficients of blocks for explaining a conventional operation.

[Explanation of symbols]

 3 Sorter 70 Classifier 60 Maximum class detector 80 Holder

Claims (1)

[Claims] 1. An orthogonal transforming means for orthogonally transforming a block-shaped signal, a coefficient for encoding the transform coefficient orthogonally transformed by the orthogonal transforming means for each block, and a rearranging means for rearranging, and the transforming means. Class dividing means for dividing the coefficient into predetermined classes according to a plurality of threshold values, and maximum class detecting means for detecting, for each block, the maximum class having the maximum value of the conversion coefficient among the classes obtained by the class dividing means, An orthogonal transform coding device, characterized in that the operations of the classifying means and the maximum class detecting means are stopped after the maximum class is detected.