JPH05328139A - Device for picture encoder and method therefor - Google Patents

Abstract

PURPOSE:To efficiently perform coding without omitting high frequency components and unnecessarily degradating quantization characteristic at the time of quantization. CONSTITUTION:A quantization DCT coefficient including the DC component from a direct current component to a high frequency component is inputted to a zero detecting part 301 and a counter 303, and a quantization threshold is incremented by a quantization threshold updating part 302 when coefficient '0' is detected by a zero detecting part 301, the quantization threshold is initialized when the coefficient other than '0' is detected. Meanwhile, the counter 303 counts the quantization coefficient including coefficient '0', and in a threshold upper limit setting part 304 the upper limit of the threshold is updated in the quantization threshold update part 302 in accordance with the threshold upper limit control signal from a picture coding control part 214 and the output of the counter 303.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image coding apparatus and method for quantizing transform coefficients obtained by orthogonally transforming an input image in pixel block units.

[0002]

2. Description of the Related Art Conventionally, when an input image is orthogonally transformed in pixel block units and its transform coefficient is quantized, the quantization threshold is updated as shown in FIG. here,
An example will be given in which the quantization step size is g, the increment amount at the time of updating is “1”, and the maximum threshold value is 1.5 g with respect to the quantization step size g. Also, the initial value of the quantization threshold is set equal to the quantization step size.

When the DCT (discrete cosine transform) coefficient input by multiplying the DC component by the high frequency component is input with data equal to or less than the threshold value, the threshold value is incremented by "1". However, the threshold has an upper limit (1.5 g) and is not updated beyond that. When the DCT coefficient data that exceeds the threshold is input, the threshold returns to the initial value (quantization step size g). The above operation promotes the generation of zero,
It is configured to easily cause “zero run” and suppress the amount of data to be transmitted.

Further, in general, high frequency components are very small in the spectrum of an image and are concentrated in the low frequency region, so that the high frequency components in the DCT coefficient are discarded to reduce the data amount, It is also done not to transmit.

[0005]

However, when the DCT process is performed on the actual image as it is (INTRA mode), when the difference between frames is DCT processed (INTER mode), the power of the DCT coefficient is in the high frequency region. Many are distributed. Further, the high frequency component is essentially important, and if the high frequency region of the DCT coefficient is cut off, a fine change may be lost, resulting in a blurred image.

If the DCT coefficient is to be quantized from the direct current component to the high frequency component, the amount of data inevitably increases. As a countermeasure, if the quantization step size is set to a large value, deterioration of the image quality as a whole cannot be avoided, and if the upper limit of the threshold value is set carelessly, important data components will be lost. There were times when it ended up.

The present invention has been made in order to solve the above problems, and can perform efficient coding without discarding high frequency components and unnecessarily lowering the quantization characteristics during quantization. An object is to provide an image encoding device and method.

[0008]

In order to achieve the above-mentioned object, the structure of the present invention is such that image coding using a variable quantization threshold when quantizing transform coefficients obtained by orthogonally transforming an input image in pixel block units. The apparatus is characterized by comprising setting means for variably setting the upper limit of the quantization threshold and quantization means for performing quantization with the quantization threshold set by the setting means.

Further preferably, the setting means variably sets the upper limit value of the quantization threshold according to the frequency domain.

[0010]

In the above configuration, when the transform coefficient obtained by orthogonally transforming the input image in pixel block units is quantized, the upper limit of the quantization threshold is variably set, and the quantization is performed by the quantization threshold. To do.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT A preferred embodiment of the present invention will be described in detail below with reference to the drawings.

FIG. 1 is a schematic block diagram showing the structure of an image communication telephone according to an embodiment. In the figure, 101
Is one of the image input means of this apparatus, and is a camera input unit for inputting a self-portrait and the like. Reference numeral 102 denotes one of the image input means of the present apparatus, which is a document camera input unit for inputting images such as drawings, maps, and documents. A display unit 103 displays an input image from the camera input unit 101 or the document camera input unit 102, an image received from a partner, an operation screen, or the like. An image input interface unit 104 performs switching processing of image input means according to an instruction from a system control unit 113 described later. An image output interface unit 105 performs switching processing of image output means. 106
Is an image editing unit for performing picture-in-picture processing, screen freezing processing, and the like according to an instruction from the system control unit 113 described later. Reference numeral 107 is an image encoding / decoding unit that performs transmission image signal encoding processing and reception image signal encoding processing, 107a is an image encoding unit that performs encoding processing, and 107b is an image decoding that performs decoding processing. It is a chemical department.

Reference numeral 108 denotes a handset unit which is a voice input / output unit of this apparatus. Reference numeral 109 denotes an eco-cancellation process, a tone generation process such as a dial tone, a ring tone, a business tone, a ring tone, and a switching process of voice input / output means according to an instruction from a system control unit 113 described later. This is a voice input / output interface section. Reference numeral 110 denotes a voice encoding / decoding unit that performs transmission voice signal encoding processing and reception voice signal decoding processing according to an instruction from the system control unit 113. Reference numeral 110a denotes a voice encoding unit that performs encoding processing.
Reference numeral 10b is a voice decoding unit that performs a decoding process. 111
Is an operation unit including a keyboard, a touch panel, etc. used for inputting control information for performing overall control of the apparatus.
Reference numeral 112 multiplexes the image signal from the image encoding / decoding unit 107, the audio signal from the audio encoding / decoding unit 110, and the control signal from the system control unit 113 on a transmission frame basis, and also receives a received frame. A demultiplexing / multiplexing unit that separates an image signal, an audio signal, and a control signal and transfers them to the above-mentioned units. A system 113 includes a CPU, a ROM, a RAM, an auxiliary storage device, and the like, monitors the state of each unit, controls the entire apparatus, creates an operation / display screen according to the state, and executes an application program. It is a control unit. A line interface unit 114 controls the line according to the ISDN user network interface. And
Reference numeral 115 is a communication line connected to an ISDN (not shown).

An outline of the operation of the present apparatus having the above configuration will be described below.

First, the input image from the camera input unit 101 or the document camera input unit 102 is passed through the image input interface unit 104 and the image editing unit 106, and then the image encoding unit 1
It is input to 07a. The input voice from the handset unit 108 is input to the voice encoding unit 110a via the voice input / output interface unit 109. Image coding unit 107a
The input image encoded by the input audio and the input audio encoded by the audio encoding unit 110a are multiplexed by the demultiplexing and multiplexing unit 112, and the communication line 115 is passed through the line interface unit 114.
Sent to.

On the other hand, the received signal from the communication line 115 is
Demultiplexing and multiplexing unit 112 via line interface unit 114
Are separated into an image signal and an audio signal by the image decoding unit 1
07b is input to the speech decoding unit 110b. The received image encoded by the image decoding unit 107b is the image editing unit 1
06, display unit 1 via image output interface unit 105
It is displayed on 03. The received voice decoded by the voice decoding unit 110b is output to the handset unit 108 via the voice input / output interface unit 109.

Next, the image coding unit 10 in this embodiment.
The detailed configuration of 7a will be described below with reference to FIG.

In FIG. 2, reference numeral 201 denotes an encoding mode selection unit, which receives input pixel value data and a difference value (prediction error) between the input pixel value data and the prediction value data from the reference frame (previous frame). The inter-frame difference value (prediction error value) is encoded with respect to the pixel value data input in macroblock units according to the energy comparison result between the two and the instruction of the image encoding control unit 214 (INTER mode),
It is selected whether the input pixel value data is encoded as it is (INTRA mode). When the INTRA mode is selected, the input pixel value is output, and when the INTER mode is selected, the prediction error value is output. A DCT (discrete cosine transform) unit 202 performs a DCT process, which is a kind of orthogonal transform, on the input pixel value data. A quantizer 203 selects a quantization step size according to an instruction from the image coding controller 214, and quantizes the input DCT coefficient data. A quantization threshold value setting unit 204 sets a quantization threshold value for increasing the zero run in the quantized output from the quantization unit 203. Here, the upper limit of the quantization threshold is set by an instruction from the image coding control unit 214.

Reference numeral 205 denotes a variable length coding unit, which performs variable length coding on the quantized DCT coefficient data. Two
Reference numeral 06 is a transmission buffer unit. An error correction coding unit 207 performs coding for 2-bit random error correction and 6-bit burst error correction. A dequantization unit 208 dequantizes the quantized DCT coefficient data. An IDCT (Inverse Discrete Cosine Transform) unit 209 performs IDCT processing on the input DCT coefficient data, and adds a predicted value to the data processed in the INTER mode. A frame memory unit 210 is a frame memory for motion-compensated inter-frame prediction. A motion vector detection unit 211 detects a motion vector for the previous frame of the processing macroblock of the current frame. Reference numeral 212 denotes a motion compensation unit that selects pixel value data of the macroblock unit in the previous frame according to the motion vector detected by the motion vector detection unit 211. A filter unit 213 is a low-pass filter for a block for which motion compensation interframe prediction is performed. Reference numeral 214 denotes an image encoding control unit, which transmits the image quality signal from the system control unit 113 and the transmission buffer unit 2.
Based on the buffer storage amount of 06, various coding controls such as coding mode selection, quantization step size selection, variable threshold control, significant block determination, and frame skipping (frame skip) are performed.

The data input to the coding mode selection unit 201 of the image coding unit 107a is CCITT Recommendation H.264.
Common Intermediate Format (CIF or QCI)
F).

3 to 6 are views showing the structure of the CIF format. As shown in FIG. 3, in a CIF format image, one frame is GOB (group of
The block is composed of 12 blocks called "blocks", and one GOB is composed of 33 blocks called macroblocks shown in FIG. Further, one macroblock includes six blocks (4
One Y block, Cb block, and Cr block), and one block is 8 pixels × 8 as shown in FIG.
It is composed of lines. Here, the encoding process is performed in the order of GOBs 1 to 12 shown in FIG.
In the OB, the macro blocks 1 to 33 shown in FIG. 4 are arranged in order, and in each macro block, the Y (luminance) blocks 1 to 4, Cb (color difference) blocks, and Cr (color difference) blocks are arranged in this order. Done.

In the QCIF format, as shown in FIG. 7, the number of pixels and the number of lines in the CIF format are each 1
/ 2. Coded transmission is performed in GOB units, motion compensation, quantization step size and coding mode selection are performed in macroblock units, and DCT (discrete cosine transform) and filter processing are performed in block units.

The image coding unit 307a having the above configuration
The operation of will be described below.

First, pixel value data and prediction value data in macroblock units are input to the coding mode selection unit 201. The coding mode selection unit 201 performs energy comparison between the input pixel value and the difference value (prediction error value) between the input pixel value and the prediction value in macro block units, and the result and the image coding control unit 213. The coding mode is selected in accordance with the coding mode control signal from, and the input pixel value is output to the DCT unit 202 in the INTRA mode and the prediction error value is output to the INTER mode. The DCT unit 202 performs DCT processing, which is a kind of orthogonal transformation, on the pixel value data from the coding mode selection unit 201 in block units, and outputs DCT coefficient data to the quantization unit 203. Here, the DCT unit 202
As shown in FIG. 8, the output is an example of data scanned in zigzag from the upper left (DC component) to the lower right (high frequency component). In the quantization unit 203, the image coding control unit 2
The input DCT coefficient is quantized by the step size selected according to the quantization characteristic control signal from 14. At the time of quantization, a DCT coefficient below a certain threshold is truncate to zero so that the "zero run" in the quantized data string becomes long.

Next, the threshold setting unit 204 updates the threshold to a higher value when zero occurs in the quantized output. here,
The upper limit of the quantization threshold is variable, and the image coding control unit 21
It is set according to the threshold upper limit setting signal from 4. The variable-length coding unit 205 makes a significant block determination or the like based on the coding control signal from the image coding control unit 214, and the quantized DCT coefficient is CCITT recommended H.264. Variable length coding is performed according to H.261 and output to the transmission buffer unit 206. The transmission buffer unit 206 is composed of a buffer memory, buffers the variable-length coded data, outputs the variable-length coded data to the error correction coding unit 207, and at the same time, sends the buffer storage amount to the image coding control unit 214. The error correction coding unit 207 adds error correction parity for 2-bit random error correction and 6-bit burst error correction on the receiving side to the image data input from the transmission buffer unit 206. 1 shown in FIG.
Output to 2.

Further, the inverse quantizer 208 includes the quantizer 2
The DCT coefficient quantization output of No. 03 is input, and the quantization unit 20
Inverse quantization is performed using the quantization step size selected in 3, and the DCT coefficient is output. IDCT unit 2
In 09, the DCT coefficient output of the dequantization unit 208 is IDC
For the data of the macroblock processed by T and further processed in the INTER mode, the corresponding prediction value is added and output. The frame memory unit 210 has at least 2
A frame memory for motion-compensated interframe prediction composed of a frame memory for frames, which stores pixel value output of the IDCT unit 209, and at the same time is instructed by the motion compensation unit 212 for motion-compensated interframe prediction The pixel value data of the previous frame is output. Motion output detector 21
In No. 1, the pixel value data of the previous frame in the vicinity of the processing macroblock position of the current frame is read from the frame memory unit 210 as a motion vector search window and a block matching operation is performed to detect the motion vector. In the motion compensation unit 212, according to the motion vector detected by the motion vector detection unit 211, the pixel value data of the corresponding macroblock in the previous frame is stored in the frame memory unit 2
It is read from 10 and output.

The filter unit 213 is a low-pass filter for the purpose of alleviating the discontinuity at the block boundary due to the motion compensation, and performs the filtering process on the motion compensated data to obtain the predicted value data. To the coding mode selection unit 201. The image encoding control unit 214 constantly monitors the image encoding unit in general or the image quality signal and the writing mode signal from the system control unit 113, so that the transmission buffer unit 206 does not overflow. Based on the data storage amount of the transmission buffer unit 206, the quantization in the quantization unit 203 by the quantization characteristic control signal is adaptively performed according to the change or the scene change of the input image or the image quality setting of the communication person. Step size selection and variable threshold control, coding mode selection in the coding mode selection unit 201 by the coding mode control signal, significant block determination in the variable length coding unit 205 by the coding control signal, or frame drop (frame Encoding control such as skip) is performed.

Next, the coding control in the image coding control unit 214 will be described.

As described above, the encoding modes include the INTER mode for encoding the difference from the previous frame (predicted value) and the INTRA mode for encoding the difference without taking the difference. For example, in an image with little motion or change, the current frame and the previous frame are very similar to each other, so that the temporal redundancy is reduced and the motion is large by using the INTER mode in which the difference between the current frame and the previous frame is encoded. When an image or a scene is changed, the correlation between frames is small and the coding in the same frame is effective. Therefore, the INTRA mode is used, in which the coding is performed as it is without taking the difference from the previous frame.
The two are adaptively switched.

Regarding the quantization characteristic, the quantization step
If the size is set to be small, the image quality is improved, but significant data is increased, which leads to an increase in the number of transmission bits. If it is set to a large value, the data amount will decrease but the image quality will deteriorate. Therefore, it is necessary to constantly monitor the amount of data stored in the transmission buffer and set it appropriately and efficiently. Also, processing such as frame dropping (frame skipping) is performed according to the number of generated bits.

In addition, as a preprocessing of quantization,
The amount of data to be generated is reduced by promoting the generation of zero by truncating the DCT coefficient below a certain level (quantization threshold) to zero with respect to the size and facilitating the generation of "zero run". Here, the threshold value that is zero-tracked is a variable threshold value (Variable Threshold value) that is updated to a higher value due to the occurrence of zero.

Next, the operation of updating the quantization threshold and setting the upper limit of the threshold will be described.

FIG. 9 is a schematic block diagram showing the structure of the quantization threshold setting unit 204 in this embodiment. In the figure, reference numeral 301 denotes a zero detection unit, which detects zero in the output of the quantization unit 203 and instructs the quantization threshold update unit 302 in the subsequent stage to increment the threshold value. Reference numeral 302 denotes a quantization threshold update unit, which is a quantized DCT of the zero detection unit 301.
When zero is detected in the coefficient, the quantization threshold is incremented by a fixed amount. Further, even if one exceeds the threshold value and the zero detection unit 301 detects a quantized output value other than zero, the quantization threshold value is returned to the initial value. A counter 303 counts the number of quantized DCT coefficients (including zero). A threshold upper limit setting unit 304 selects and sets the upper limit of the threshold by the output of the counter 303. The quantization threshold updating unit 302 and the counter 303 have an image coding control unit 214.
It is reset to the initial value by the block synchronization signal from.

The quantization threshold setting unit 20 having the above configuration
The operation of No. 4 will be described below. First, the quantizer 20
The DCT coefficient data quantized in 3 is input to the zero detection unit 301 and the counter 303. Here, the output of the quantizer 203 is a data string scanned in zigzag from the DC component to the high frequency component, as described above. In the zero detection unit 301, when a zero coefficient is detected in the input quantized DCT coefficient data, the quantization threshold updating unit 302 in the subsequent stage is instructed to increment the quantization threshold, and when a coefficient other than zero is detected, the quantization threshold is detected. To initialize the. In the quantization threshold updating unit 302, when the preceding zero detection unit 301 detects the coefficient zero, the quantization threshold updating unit 302 increments the quantization threshold by a fixed amount. Further, even one DCT coefficient exceeding the quantization threshold is input to the quantization unit 203, and a quantization coefficient other than zero is generated, and the zero detection unit 301 in the preceding stage is generated.
When a coefficient other than zero is detected, the quantization threshold is reset to the initial value. Here, the upper limit of the quantization threshold is set by the threshold upper limit setting unit 304. Further, the counter 303 counts the quantized coefficients including the zero coefficient,
The threshold upper limit setting unit 304 in the subsequent stage is notified of the number of the DCT coefficient in the processing block shown in FIG. 8 that is being quantized. The threshold upper limit setting unit 304 includes the image coding control unit 2
In response to the threshold upper limit control signal from 14 and the output of the counter 303, the quantization threshold updating unit 302 is instructed to set the upper limit of the threshold. The threshold upper limit control signal controls whether the upper limit of the threshold is variable or fixed according to the output of the counter 303. The counter 30 is generated by the block synchronization signal in the quantization unit 203 from the image coding control unit 214.
The counter value in 3 and the threshold value in the quantization threshold value updating unit 302 are initialized. Based on the output of the quantization threshold update unit 302 and the quantization step size instruction signal from the image coding control unit 214, the quantization unit 203 receives the input DC
The T coefficient is quantized.

Next, the above-mentioned quantization threshold setting operation will be described with a specific example.

FIG. 10 is a diagram showing a specific quantization threshold updating operation. Here, it is assumed that the increment amount (update width) of the quantization threshold value is “1” and the initial value of the threshold value is equal to the quantization step size (g). The upper limit of the quantization threshold is 3 from the DC component by the output of the counter 303.
It is assumed that 1.5 g and 2 g are switched with the second coefficient as a boundary.

In the quantizing unit 203, when the DCT coefficient input from the DCT unit 202 from the DC component to the high frequency component is equal to or less than the threshold value set by the quantization threshold setting unit 204, the quantization coefficient zero is quantized. Threshold setting unit 204
Output to the zero detecting unit 301. When the zero coefficient is detected by the zero detector 301, the threshold value is incremented by "1". As for the upper limit of the threshold value, 1.5 g is used for 32 coefficients from the DC component, and 2 g is used for the subsequent 32 coefficients. When the DCT coefficient exceeding the quantization threshold is input to the quantization unit 203 and even one coefficient other than zero is input to the zero detection unit 301, the threshold is reset to the initial value (quantization step size g). When the processing of one block (64 DCT coefficients) is completed, the quantization threshold is initialized by the block synchronization signal.

By the above operation, by raising the upper limit of the quantization threshold for high frequency components, the high frequency components are not completely cut off, and only a sufficiently small DCT coefficient is truncated to zero. This effectively promotes the occurrence of "zero run" and suppresses the amount of data to be transmitted.

In this embodiment, only the case of having two kinds of threshold upper limit values was described, but it goes without saying that the present invention can be similarly applied to the case of having a plurality of variable upper limit values.

The present invention may be applied to a system composed of a plurality of devices or an apparatus composed of a single device.

Further, it goes without saying that the present invention can be applied to the case where it is achieved by supplying a program to a system or an apparatus.

[0042]

As described in detail above, according to the present invention, there is provided an image coding apparatus which uses a variable quantization threshold when quantizing a transform coefficient obtained by orthogonally transforming an input image in pixel block units. By providing a setting unit that variably sets the upper limit of the quantization threshold and a quantization unit that performs quantization with the quantization threshold set by the setting unit, it is possible to set the adaptive quantization threshold. Therefore, efficient compression encoding of image information becomes possible. Furthermore, by setting the upper limit of the quantization threshold variable according to the frequency domain, efficient encoding according to frequency components can be performed without discarding important high frequency components or unnecessarily lowering the quantization characteristics. Will be possible.

[Brief description of drawings]

FIG. 1 is a schematic block diagram showing a configuration of an image communication telephone according to an embodiment.

FIG. 2 is a block diagram showing a detailed configuration of an image encoding unit shown in FIG.

FIG. 3 is a diagram showing a CIF format composed of 12 GOBs.

FIG. 4 G of FIG. 3 composed of 33 macroblocks
It is a figure which shows OB.

FIG. 5 is a diagram showing the macroblock of FIG. 4, which is composed of six blocks.

6 is a diagram showing the block of FIG. 5 consisting of 8 pixels × 8 lines.

FIG. 7 is a diagram showing a structure of a QCIF format.

FIG. 8 is a diagram for explaining zigzag scanning.

9 is a schematic block diagram showing a configuration of a quantization threshold setting unit shown in FIG.

FIG. 10 is a diagram showing a quantization threshold updating operation in the present embodiment.

FIG. 11 is a diagram showing a quantization threshold update operation in a conventional example.

Claims (3)

[Claims] 1. An image coding apparatus that uses a variable quantization threshold when quantizing a transform coefficient obtained by orthogonally transforming an input image in pixel block units, and a setting means for variably setting an upper limit of the quantization threshold. An image coding apparatus, comprising: a quantizing unit that performs quantization according to the quantization threshold set by the setting unit. 2. The image coding apparatus according to claim 1, wherein the setting unit variably sets the upper limit value of the quantization threshold according to the frequency domain. 3. An image coding method that uses a variable quantization threshold when quantizing a transform coefficient obtained by orthogonally transforming an input image in pixel block units, comprising a setting step of variably setting an upper limit of the quantization threshold. And a quantization step of performing quantization with the quantization threshold set in the setting step.