JP2005176169A - Video coding method and apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To realize video coding to handle signals having the different numbers of effective bits without expanding a circuit scale. <P>SOLUTION: There is provided a video coding method or the like including steps of: receiving a video signal resulting from multiplexing a plurality of signals having different effective bit lengths; extracting each of the plurality of signals from the received video signal, performing orthogonal transform thereon and outputting transform factors; classifying the transform factors based on a predetermined threshold; converting a bit length of each classified transform factor into a predetermined bit length for each transform factor and limiting the bit length; quantizing the transform factor converted into the predetermined bit length; and coding the quantized transform factor. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

    The present invention relates to video encoding used for high-efficiency encoding of video signals.

  With the digitization of video signals, a technique for recording or transmitting video signals with high-efficiency encoding has become important (see, for example, Patent Document 1). As one of the high-efficiency encoding techniques, there is orthogonal transform coding that reduces the amount of data by a method such as blocking a video signal and performing orthogonal transform, and quantizing a high frequency component with a larger quantization width. In addition, by weighting the quantization width of each block using the difference in amplitude of frequency components when orthogonally transforming the block, a complex picture that is not visually noticeable even when block noise occurs, that is, a large amplitude There is a technique for positively quantizing a block including a frequency component. Thereby, the block of the flat pattern where block noise is conspicuous can be protected.

  FIG. 4 is a block diagram showing a configuration of a conventional video encoding device 400. As shown in FIG. The video encoding device 400 weights the quantization width for each block and performs orthogonal transform encoding. The video encoding device 400 includes an input terminal 401, an orthogonal transform unit 402, a maximum value detection unit 403, a classification unit 404, an initial quantization unit 405, a quantization selection unit 406, a quantization unit 407, An encoding unit 408 and an output terminal 409 are provided.

Hereinafter, the operation of each component will be described. The orthogonal transform unit 402 performs orthogonal transform on the video signal input from the input terminal 401 and outputs an m-bit length transform coefficient indicating a frequency component. The maximum value detection unit 403 detects the maximum value Am of the absolute value amplitude of the conversion coefficient. The classification unit 404 performs classification (classification) by determining which region the maximum value Am belongs to among a plurality of regions divided by a predetermined threshold. For example, when three threshold values A0, A1, and A2 (A0 <A1 <A2) are set, the maximum value is shown in Table 1 below.
Belongs to one of the four classes 0 to 3.

  By setting the threshold value A2 to a value that can be represented by (m−1) bits, all transform coefficients belonging to a block satisfying Am <A2 can be represented by (m−1) bits. Therefore, the most significant bit of m bits is sliced, that is, truncated, and converted to a bit length of (m−1) bits. In addition, blocks classified as class 3 are high-amplitude frequency components and block noise is inconspicuous. Therefore, even if quantization is performed with a larger quantization width than other classes, the visual effect is not affected. Few. Therefore, it is possible to quantize from m bits to (m−1) bits in advance. FIG. 5 is a diagram illustrating how the transform coefficient is limited to a predetermined number of bits. The example of FIG. 5 shows a case where the transform coefficient is composed of a 1-bit code bit (symbol s in the figure) and a 9-bit data part (a part in which numerical values are shown in the figure). The “most significant bit” described above represents the most significant bit of the 9-bit data portion. The initial quantization unit 405 (FIG. 4) selects the sign bit and the upper 8 bits of the data portion from the transform coefficient in the case of class 3, and the other classes (class 0, class 1 and class 2). In this case, by selecting the sign bit and the lower 8 bits of the data part, the transform coefficient is limited to a predetermined bit length.

  Referring to FIG. 4 again, quantization selection section 406 determines a quantization width such that the encoded data is equal to or smaller than a predetermined value. At this time, according to the classification information output from the classification unit 404, an offset is given to the quantization width selected according to the class as shown in Table 2 below. This is because the block having a larger amplitude of the frequency component performs quantization with a larger quantization width.

  The numerical values in Table 2 indicate divisors when the quantization unit 407 performs quantization. The class 3 block has already been quantized by the initial quantization unit 405. Therefore, the total quantization width of the initial quantization unit 405 and the quantization unit 407 is further increased by the class 3 divisor as shown in Table 3 below.

  Here, for example, when the total amount of data after coding of a class 0 block and a class 1 block is equal to or less than a predetermined amount, the class 1 block is divided by a divisor twice as large as the class 0 block. The amount of data allocated to a class 1 block having a large amplitude can be suppressed, and the amount of data allocated to a class 0 block having a small frequency component amplitude can be protected.

  The quantization unit 407 quantizes the transform coefficient limited to (m−1) bits with the quantization width determined by the quantization selection unit 406. The output from the quantization unit 407 is converted into a code word by the encoding unit 408 and output from the output terminal 409.

In this way, by limiting the transform coefficient to (m−1) bits by classification, the bit accuracy of the quantization selection unit 406, the quantization unit 407, and the encoding unit 408 corresponds to (m−1) bits. This can be done and the circuit scale can be reduced. Further, extra quantization is not performed on a block having a small frequency component amplitude, and encoding can be performed with (m−1) -bit circuit resources without degrading the calculation accuracy.
JP 2003-23625 A

  However, the conventional video encoding device 400 cannot simultaneously handle a plurality of input video signals having different effective bit numbers. As a result, it is necessary to have double circuits having similar functions and switch the encoded data. This may lead to an increase in circuit scale and power consumption.

  On the other hand, if the circuit is simply shared in order to reduce the circuit scale, the quantization width of each block may be inappropriately weighted. Specifically, the signal obtained by taking the difference between the reference video and the input video that are temporally different from each other has a large amplitude of the frequency component when the block is orthogonally transformed if the original image is a flat image. However, even if block noise occurs in a block having a large frequency component amplitude, the block cannot be said to be a visually inconspicuous picture.

  An object of the present invention is to realize video coding capable of handling signals having different effective bit numbers without increasing the circuit scale. Another object is to share a circuit while realizing video coding in which block noise is hardly noticeable visually.

  A video encoding method according to the present invention includes a step of receiving a video signal in which a plurality of signals having different effective bit lengths are multiplexed, taking out each of the plurality of signals from the received video signal, and performing orthogonal transform, A step of outputting a transform coefficient; a step of classifying the transform coefficient based on a predetermined threshold; and a bit length by converting the bit length of the transform coefficient into a predetermined bit length for each of the classified transform coefficients , The step of quantizing the transform coefficient converted to the predetermined bit length, and the step of encoding the quantized transform coefficient, thereby achieving the above object.

The step of converting may be a step of quantizing the bit length of the conversion coefficient with a predetermined quantization width and converting it to a predetermined bit length.
The classifying step may be performed based on a predetermined threshold and a maximum absolute value of the conversion coefficient.
The predetermined threshold may be provided for each of the plurality of signals having different effective bit lengths, and the predetermined quantization width may be different for each of the plurality of signals having different effective bit lengths.

The classifying step classifies the conversion coefficients of the first signal with respect to all the conversion coefficients, and the conversion coefficients of the second signal include: It may be a step of performing classification for AC value conversion coefficients that do not include DC values.
The video signal in which a plurality of signals are multiplexed includes a first signal that takes a difference between a reference video that is temporally mixed and an input video, and a second signal that represents the input video, and is converted. The step substantially matches the quantization width for each classification of the transform coefficient for the first signal, and classifies the transform coefficient having a large absolute value for the second signal. The quantization width may be increased.

  The quantization step performs quantization for all the transform coefficients of the first signal, and transforms the transform coefficient of the second signal with respect to the transform coefficient of the second signal. Of these steps, quantization may be performed on a conversion coefficient of an AC value that does not include a DC value.

  The video encoding apparatus according to the present invention includes an input terminal for receiving a video signal in which a plurality of signals having different effective bit lengths are multiplexed, and taking out each of the plurality of signals from the video signal received at the input terminal. For each of the transform coefficients classified by the classifying unit, the classifying unit that classifies the transform coefficients output from the orthogonal transform unit based on a predetermined threshold, and the orthogonal transform unit that converts and outputs each transform coefficient An initial quantization unit that converts the bit length of the transform coefficient to a predetermined bit length to limit the bit length, and a quantization unit that quantizes the transform coefficient converted to the predetermined bit length by the initial quantization unit And an encoding unit that encodes the transform coefficient quantized by the quantization unit, thereby achieving the above object.

The initial quantization unit may quantize the bit length of the transform coefficient with a predetermined quantization width and convert it to a predetermined bit length.
A maximum value detecting unit that detects a maximum value of the absolute value of the transform coefficient output from the orthogonal transform unit; the classifying unit includes a predetermined threshold and the absolute value of the transform coefficient detected by the maximum value detecting unit; The conversion coefficient may be classified based on the maximum value of.

The predetermined threshold may be provided for each of the plurality of signals having different effective bit lengths, and the predetermined quantization width may be different for each of the plurality of signals having different effective bit lengths.
The classification unit includes: a first classification unit that performs classification on all of the conversion coefficients of the first signal; and a conversion coefficient of the conversion coefficient of the second signal. Of these, a second classification unit that performs classification on an AC value conversion coefficient that does not include a DC value may be included.

  The video signal in which a plurality of signals are multiplexed includes a first signal that takes a difference between a reference video that is temporally mixed and an input video, and a second signal that represents the input video, and an initial quantization The unit substantially matches the quantization width for each classification of the transform coefficient for the first signal, and classifies the transform coefficient having a large absolute value for the second signal. The quantization width may be increased.

  The quantization unit quantizes all the transform coefficients for the transform coefficient of the first signal, and the DC value of the transform coefficients for the transform coefficient of the second signal. Quantization may be performed on a conversion coefficient of an AC value that does not include the.

  According to the present invention, separate classification units are provided for video signals having different effective bit lengths, and the bit length of the transform coefficient is limited to a predetermined bit length. As a result, the subsequent processing falls within the predetermined bit length, and the circuit and the like can be shared. That is, video signals having different effective bit lengths can be processed without increasing the circuit scale and power consumption, which is very practical.

  In addition, according to the present invention, since the class information is used only for bit length restriction on the signal obtained by taking the difference between the reference video and the input video, the total quantization of the quantization performed by the initial quantization unit and the quantization unit The quantization width of the quantization unit can be weighted so that the widths are almost the same. As a result, image quality degradation can be suppressed by selecting a quantization that matches the characteristics of the input signal, and the subsequent processing circuits and the like can be shared. Therefore, it is possible to realize processing with little increase in circuit scale and power consumption.

  Further, according to the present invention, in the case of a signal having a difference, the maximum value of the absolute value is obtained for the transform coefficient including the DC value, classification is performed, and encoding is performed. As a result, the amount of data used for the DC value can be reduced and efficient overall coding can be performed, and the image quality can be improved.

Embodiments of the present invention will be described below with reference to the drawings.
(Embodiment 1)
FIG. 1 is a block diagram showing a configuration of video encoding apparatus 100 in the first embodiment. The video encoding apparatus 100 includes an input terminal 101, an orthogonal transform unit 102, a maximum value detection unit 103, a first classification unit 104, an initial quantization unit 105, a quantization selection unit 106, and a quantization unit 107. A coding unit 108, an output terminal 109, a second classification unit 110, and a switch 111.

  Hereinafter, the operation of each component of the video encoding device 100 will be described. In the present embodiment, different classification is performed for each signal having a different number of effective bits, and initial quantization is performed from the input signal and the class information, and control is performed to limit the bit length of the transform coefficient to a predetermined bit length. In the following, for the sake of simplicity, the signal input from the input terminal 101 is a video signal obtained by multiplexing two types of signals that are different in effective bit length by 1 bit.

  The orthogonal transform unit 102 performs orthogonal transform on the video signal input from the input terminal 101. More specifically, each of the two multiplexed signals constituting the video signal is taken out and orthogonally transformed. Then, corresponding conversion coefficients having effective bit lengths indicating frequency components of m bits and (m + 1) bits are output. The maximum value detection unit 103 detects the maximum value Am of the absolute value amplitude of the conversion coefficient. The first classification unit 104 sets three threshold values A0, A1, and A2 (A0 <A1 <A2) shown in Table 4 below. The second classification unit 110 sets three new threshold values B0, B1, and B2 (B0 <B1 <B2) shown in Table 5.

  Each of the first classification unit 104 and the second classification unit 110 performs classification (classification) by determining which region the maximum value Am belongs to among a plurality of ranges divided by a predetermined threshold. That is, the first classification unit 104 and the second classification unit 110 perform two types of classification. The switch 111 selects the output of the first classifying unit 104 when the effective bit length is m bits, and selects the output of the second classifying unit 110 when the effective bit length is (m + 1) bits. Output.

  In the first embodiment, for a block having an effective bit length of m bits, the threshold value A2 is set to a value that can be represented by (m−1) bits. For class 3 blocks, quantization processing to (m−1) bits is performed. In addition, for the blocks of classes other than class 3 (class 0, class 1 and class 2), the initial quantization unit 105 performs upper bit slice processing. Thereby, the input to the quantization unit 107 can be limited to (m−1) bits.

For a block having an effective bit length of (m + 1) bits, the threshold value B1 is set to a value that can be expressed by (m−1) bits, and the threshold value B2 is set to a value that can be expressed by m bits. Then, a 2-bit quantization process is performed on the class 3 block. The quantization processing for 2 bits here means that a class 3 block is divided by 4 (= 2 ² ). In addition, for a class 2 block, quantization processing for 1 bit and slice processing for the most significant bit are performed. The quantization processing for 1 bit represents dividing a class 2 block by 2 (= 2 ¹ ). For classes other than class 2 and class 3 (class 0 and class 1), the initial quantization unit 105 performs upper 2-bit slice processing. Thereby, the input to the quantization unit 107 can be limited to (m−1) bits, that is, converted.

  FIG. 2 is a diagram illustrating how the transform coefficient is limited to a predetermined number of bits. In the example of FIG. 2, the transform coefficient is composed of a 1-bit code bit (symbol s in the figure) and a 10-bit data part (a part in which numerical values are shown in the figure). Numerical values 0 to 9 of the 10-bit data portion represent the 0th to 9th bits. In the case of class 3, the initial quantization unit 105 (FIG. 1) selects the sign bit and the upper 8 bits (second to ninth bits) of the data part in the case of class 3. In the case of class 2, 8 bits (first to eighth bits) excluding the most significant bit and the least significant bit of the data bit are selected. For the other classes (class 0 and class 1), the sign bit and the lower 8 bits (0th to 7th bits) of the data part are selected. Thereby, the conversion coefficient can be easily converted into a predetermined bit length and limited.

  Thus, in the first embodiment, the first classification unit 104 and the second classification unit 110 are switched according to the difference in the bit length of the input signal, and the output of the initial quantization unit 105 is set to a predetermined bit length (here (( m-1) bit). As a result, the subsequent processing (processing of the quantization selection unit 106, the quantization unit 107, and the encoding unit 108) is executed with a bit accuracy of (m−1) bits using components that have been conventionally used. can do. Specifically, the quantization selection unit 406 determines a quantization width such that the encoded data is equal to or less than a predetermined value. At this time, according to the classification information obtained via the switch 111, an offset is given to the quantization width selected according to the class as shown in Table 6 below. This is because the block having a larger amplitude of the frequency component performs quantization with a larger quantization width.

  The numerical values in Table 6 indicate divisors when the quantization unit 107 performs quantization. The class 3 block has already been quantized by the initial quantization unit 105. Therefore, the total quantization width of the initial quantization unit 105 and the quantization unit 107 is further increased by the class 3 divisor as shown in Table 7 below.

Here, for example, when the total amount of data after coding of a class 0 block and a class 1 block is equal to or less than a predetermined amount, the class 1 block is divided by a divisor twice as large as the class 0 block. The amount of data allocated to a class 1 block having a large amplitude can be suppressed, and the amount of data allocated to a class 0 block having a small frequency component amplitude can be protected.
The quantization unit 107 quantizes the transform coefficient limited to (m−1) bits with the quantization width determined by the quantization selection unit 106. The output from the quantization unit 107 is converted into a code word by the encoding unit 108 and output from the output terminal 109.

  As described above, even when video signals having different bit lengths are input, a plurality of classification units are provided and selectively used, and the quantization processing step of the initial quantization unit 105 is changed. As a result, the input signal of the quantizing unit 107 can be limited to a predetermined bit length, and can be encoded without increasing the scale of the circuit and without reducing the accuracy of the block having a small frequency component amplitude. .

  Here, the threshold value of the second classification unit 110 is set to three, which is the same as the threshold value of the first classification unit 104, but may be four or more, or a different number such as two. Moreover, the process of the 1st classification | category part 104 and the 2nd classification | category part 110 can be shared by setting the threshold value A2 and the threshold value B1 to an equal value.

(Embodiment 2)
In the second embodiment, when the video signal includes a signal obtained by taking a difference between the reference video and the input video that are temporally different from each other and an input video signal, the conversion coefficient is obtained in the case of a signal that has not taken the difference. For a class with a larger amplitude width, the quantization width is increased, and in the case of a signal having a difference, the quantization width for each class is made substantially coincident and used only for limiting the bit length. Further, in the case of a signal having a difference, efficient encoding is performed by obtaining the maximum absolute value of the transform coefficient including the DC value and performing quantization.

  FIG. 3 is a block diagram showing a configuration of video encoding apparatus 300 in the second embodiment. The video encoding device 300 includes an input terminal 301, an orthogonal transform unit 302, a maximum value detection unit 303, a first classification unit 304, an initial quantization unit 305, a quantization selection unit 306, and a quantization unit 307. An encoding unit 308, an output terminal 309, a second classification unit 310, a switch 111, and a weighting unit 312.

Hereinafter, the operation of each component of the video encoding device 300 will be described.
A processing signal is input to the input terminal 301. The processing signal refers to a signal obtained by taking a difference between the reference video and the input video (hereinafter referred to as a differential signal) when there is an appropriate reference video that changes in time with respect to the input video. In addition, when an appropriate reference video is not found and a difference is not obtained, it refers to an input video itself (hereinafter referred to as an intra signal). When the processing signal is a differential signal, it is assumed that the effective bit length is increased by 1 bit by taking the difference between the reference video and the input video.

  The orthogonal transform unit 302 performs orthogonal transform on the processing signal input from the input terminal 301 and outputs a transform coefficient. More specifically, when the processing signal is an intra signal, a conversion coefficient having an effective bit length indicating a frequency component of m bits is output. When the processed signal is a differential signal, a conversion coefficient having an effective bit length indicating a frequency component of (m + 1) bits is output.

  The maximum value detection unit 303 detects the maximum value Am of the absolute value amplitude of the transform coefficient output from the orthogonal transform unit 302. At this time, if the block being processed is a differential signal, the DC value also takes a small value, so that the maximum value is detected for all transform coefficients including the DC value as the target of the variable length code. Here, the DC value refers to a zeroth order coefficient representing a low frequency component in the conversion coefficient, and represents a direct current value. On the other hand, in the case of an intra signal, a conversion coefficient excluding a DC value, that is, an AC value is targeted. The AC value refers to a first-order or higher coefficient representing a relatively high frequency component in the conversion coefficient, and represents an AC value. The detected maximum value Am of the absolute value of the conversion coefficient is put into a condition for classifying. The first classification unit 304 sets the three threshold values A0, A1, and A2 (A0 <A1 <A2) shown in Table 4 above. The second classification unit 310 sets two new threshold values C0 and C1 (C0 <C1) shown in Table 8 below.

  Each of the first classification unit 304 and the second classification unit 310 performs classification (classification) by determining which region the maximum value Am belongs to among a plurality of regions divided by a predetermined threshold. The switch 311 selects the output of the first classifying unit 304 when the block being processed is an intra signal, and selects the output of the second classifying unit 310 when the block is a differential signal. Output.

  For a block in which the input signal is an intra signal, the threshold A2 is set to a value that can be represented by (m−1) bits. For class 3 blocks, quantization processing to (m−1) bits is performed. In addition, for the blocks of classes other than class 3 (class 0, class 1 and class 2), the initial quantization unit 305 performs higher-order bit slice processing. Thereby, the input to the quantization unit 307 can be limited to (m−1) bits.

For a block in which the input signal is a differential signal, the threshold value C0 is set to a value that can be represented by (m−1) bits, and the threshold value C1 is set to a value that can be represented by m bits. Then, the initial quantization unit 305 performs the following processing. That is, a 2-bit quantization process is performed on the class 2 block. For a class 1 block, quantization processing for 1 bit and slice processing for the most significant bit are performed. For class 0, the upper 2 bits are sliced. Thereby, the input to the quantization unit 307 can be limited to (m−1) bits.

  A block having a differential signal as an input signal has a feature that block noise including a frequency component having a large amplitude is not conspicuous. Therefore, quantization is performed by assigning a quantization width having an offset to each class shown in Table 6. be able to. On the other hand, a weighting unit 112 is provided for a block whose input signal is a differential signal. The weighting unit 112 sets the quantization width shown in Table 9 so as to cancel the quantization for limiting the bit length performed by the initial quantization unit 305.

  The reason for providing such a weighting unit 112 is that, even in a block including a frequency component having a large amplitude, a processed signal before taking a difference may represent a flat image that is a conspicuous pattern when block noise occurs. Because there is. Thereby, the total quantization width of the quantization performed by the initial quantization unit 305 and the quantization unit 307 becomes a value shown in Table 10, and the total quantization widths of the respective classes can be substantially matched.

  As described above, in the second embodiment, when the input signal is a differential signal, a weighting unit 312 is provided that selectively uses a plurality of classification units and switches the quantization width of the quantization selection unit 306. It was. Thereby, the input signal of the quantization part 307 can be restrict | limited to predetermined bit length, and the quantization suitable for the kind of input signal can be performed.

  In addition, here, for the sake of simplicity, the case where the inside of a block is divided by one kind of quantization width has been described, but when weighting is performed to quantize the high frequency component of the transform coefficient after orthogonal transformation with a larger quantization width Can also be adapted. In addition, although the switch 311 is provided before the initial quantization unit 305, the process up to the initial quantization may be performed using another circuit, and then the switch may be provided. Further, with respect to the processing for the DC value in the maximum value detection unit 303, the maximum value in the transform coefficient that does not include the DC value is detected in both the intra signal and the difference signal. You may provide the comparison part which performs magnitude comparison with a value.

1 is a block diagram showing a configuration of a video encoding device 100 according to Embodiment 1. FIG. It is a figure which shows a mode that a conversion factor is restrict | limited to a predetermined number of bits. 6 is a block diagram showing a configuration of a video encoding device 300 according to Embodiment 2. FIG. It is a block diagram which shows the structure of the conventional video coding apparatus 400. FIG. It is a figure which shows a mode that a conversion factor is restrict | limited to a predetermined number of bits.

Explanation of symbols

DESCRIPTION OF SYMBOLS 101 Input terminal 102 Orthogonal transformation part 103 Maximum value detection part 104 1st classification | category part 105 Initial quantization part 106 Quantization selection part 107 Quantization part 108 Encoding part 109 Output terminal 110 2nd classification | category part 111 Switch

Claims (14)

Receiving a video signal in which a plurality of signals having different effective bit lengths are multiplexed;
Extracting each of the plurality of signals from the received video signal, performing orthogonal transform, and outputting each transform coefficient;
Classifying the transform coefficients based on a predetermined threshold;
For each of the classified transform coefficients, converting the bit length of the transform coefficient to a predetermined bit length to limit the bit length;
Quantizing the transform coefficient transformed to the predetermined bit length;
And a method of encoding the quantized transform coefficient.   The video encoding method according to claim 1, wherein the converting step is a step of quantizing the bit length of the conversion coefficient with a predetermined quantization width and converting the quantized bit length into a predetermined bit length.   The video encoding method according to claim 2, wherein the classifying step is performed based on a predetermined threshold value and a maximum absolute value of the transform coefficient.   The video encoding according to claim 3, wherein the predetermined threshold is provided for each of the plurality of signals having different effective bit lengths, and the predetermined quantization width is different for each of the plurality of signals having different effective bit lengths. Method.   The classifying step classifies the conversion coefficients of the first signal with respect to all the conversion coefficients, and the conversion coefficients of the second signal include: The video encoding method according to claim 4, wherein classification is performed on a conversion coefficient of an AC value that does not include a DC value. The video signal in which a plurality of signals are multiplexed includes a first signal that takes a difference between a reference video that is temporally mixed and an input video, and a second signal that represents the input video,
The transforming step substantially matches the quantization width for each classification of the transform coefficients for the first signal, and the transform with a large absolute value for the second signal. The video encoding method according to claim 5, wherein the quantization width is increased as the coefficients are classified.   The quantization step performs quantization for all the transform coefficients of the first signal, and transforms the transform coefficient of the second signal with respect to the transform coefficient of the second signal. 6. The video encoding method according to claim 5, wherein the video encoding method is a step of performing quantization on a conversion coefficient of an AC value not including a DC value. An input terminal for receiving a video signal in which a plurality of signals having different effective bit lengths are multiplexed;
From the video signal received at the input terminal, each of the plurality of signals is extracted and orthogonally transformed, and an orthogonal transformation unit that outputs each transformation coefficient;
A classifying unit that classifies the transform coefficient output from the orthogonal transform unit based on a predetermined threshold;
For each transform coefficient classified by the classification unit, an initial quantization unit that converts the bit length of the transform coefficient to a predetermined bit length and limits the bit length;
A quantization unit that quantizes the transform coefficient converted to the predetermined bit length by the initial quantization unit;
And a coding unit that codes the transform coefficient quantized by the quantization unit.   The video encoding device according to claim 8, wherein the initial quantization unit quantizes the bit length of the transform coefficient with a predetermined quantization width and converts the quantized bit length to a predetermined bit length. A maximum value detecting unit for detecting a maximum value of the absolute value of the converting unit output from the orthogonal transforming unit;
The video encoding device according to claim 9, wherein the classification unit classifies the transform coefficients based on a predetermined threshold and a maximum absolute value of the transform coefficients detected by the maximum value detection unit.   The video encoding according to claim 10, wherein the predetermined threshold is provided for each of the plurality of signals having different effective bit lengths, and the predetermined quantization width is different for each of the plurality of signals having different effective bit lengths. apparatus.   The classification unit includes: a first classification unit that performs classification on all of the conversion coefficients of the first signal; and a conversion coefficient of the conversion coefficient of the second signal. The video encoding apparatus according to claim 11, further comprising: a second classifying unit that classifies AC coefficient conversion coefficients not including a DC value as a target. The video signal in which a plurality of signals are multiplexed includes a first signal that takes a difference between a reference video that is temporally mixed and an input video, and a second signal that represents the input video,
The initial quantization unit substantially matches the quantization width for each classification of the transform coefficients with respect to the first signal, and the transform with a large maximum absolute value for the second signal. The video encoding device according to claim 12, wherein the quantization width is increased as the coefficients are classified. The quantization unit quantizes all the transform coefficients for the transform coefficient of the first signal, and the DC value of the transform coefficients for the transform coefficient of the second signal. The video encoding apparatus according to claim 12, wherein quantization is performed on a conversion coefficient of an AC value that does not include an image.

Images