JPH0730889A - Picture data encoder - Google Patents

Abstract

PURPOSE:To reduce deterioration in picture quality by measuring a generated code quantity for each block and controlling a round-off range of picture information based on the result of measurement thereby enabling the encoding and decoding of picture data. CONSTITUTION:A DCT coefficient as a result of discrete cosine transformation at an orthogonal transformation device 3 is quantized by a quantizer 4, subject to variable length coding at a variable length encoder 5 and outputted to a storage medium of picture data. In this case, a code quantity measurement device 6 measures the code quantity of data subject to variable length coding and gives the result of measurement to a coefficient transformation section 7. The coefficient transformation section 7 compares each picture code quantity sent from the code quantity measurement device 6 with a predetermined code quantity latched in an object setting section 8. Then a coefficient transformation section 7 corrects the code quantity increasingly when the measured code quantity is lower and corrects the code quantity decreasingly when the measured code quantity is in excess.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image data coding apparatus for coding image data in pixel block units,
Particularly, the present invention relates to an image data encoding device suitable for a still image or moving image storage device or a transmission device.

[0002]

2. Description of the Related Art Standardization of image data compression methods such as JPEG and MPEG has been advanced as a high-efficiency encoding method for image data of still images or moving images. In these techniques, image data that is continuously input is divided into blocks of m × m pixels, a discrete cosine transform (DCT) process is performed on each divided block, and an image is subjected to variable length coding after zigzag scanning. This is a data encoding method.

The image data coding method using the orthogonal transformation focuses on the concentration of signal energy in the low frequency component in the frequency domain, and instead of assigning a large number of coding bits to the low frequency component, the energy component This coding method aims to achieve high efficiency by reducing the number of coding bits as a whole by a method of allocating a small number of bits to a high-frequency component which has a low sensitivity and is visually insensitive.

[0004]

A considerable amount of calculation is required for the calculation of the image data of the orthogonal transform method. In particular, when handling moving images, it is necessary to encode and decode images of a plurality of frames at high speed. Therefore, the smaller the code amount of the image data, the faster the data processing becomes possible.

Further, when encoding and decoding of image data of the orthogonal transformation system is used in a color facsimile apparatus or the like which transmits the accumulated image file through a transmission line, it is possible to specify a specific value depending on the capacity of the transmission line. If it is not less than the code amount,
There was a problem that it could not be decrypted in real time. Further, when storing image data in a storage device such as an optical disk device or an IC memory card, it is necessary to control the code amount of the entire image of one frame to a predetermined code amount depending on the capacity of the storage device. Was there.

Therefore, when encoding image data,
There has been a need for a technique of controlling the code amount of image data so that the bit rate is equal to or lower than the bit rate according to the processing performance of the image data processing device and the processing performance of peripheral devices.

As a conventional code amount control method, there is a method of controlling the code amount by changing the quantization step size for quantizing the DCT coefficient after orthogonal transformation. However, according to this method, the value of the quantization table has to be taken into the bit stream each time, which causes a problem in that the amount of information for the quantization table increases.

With respect to the above problems, the present invention is
It is a first object of the present invention to provide an image data coding apparatus capable of processing the coding and decoding of image data at a bit rate adapted to the processing performance of the image data processing apparatus and having little deterioration in image quality.

A second object of the present invention is to provide an image data coding apparatus capable of controlling the code amount of image data to a predetermined code amount or less.

[0010]

In order to solve the above-mentioned problems, the invention according to claim 1 divides image data that is continuously input into pixel blocks of m × m pixels, In an image data processing device for encoding image data in units of pixel blocks by orthogonal transformation and variable length coding, code amount measuring means for measuring by comparing the code amount of the pixel block with a predetermined code amount, It is characterized by control means for controlling the cut-off range of image information based on the measurement result of the code amount measuring means.

According to a second aspect of the present invention, the control means divides the image information for each block into groups of pixel information of a plurality of pixels, and controls the cut-off range of the image information for each group. Is characterized by.

Further, the invention according to claim 3 is characterized in that the control means controls the cut-off range of the image information separately for a picture for intra-frame coding and a picture for inter-frame coding. I am trying.

[0013]

In the image data coding apparatus of the orthogonal transform system, the generated code amount is measured for each block so that the code amount after compression is equal to or less than the required code amount, and the entire image is measured based on this measurement result. So that the code amount of
The code amount of the image is controlled by changing the cutoff range of the CT coefficient.

[0014]

Embodiments of the present invention will now be described in detail with reference to the drawings. FIG. 1 is a block diagram of an image data encoding apparatus to which the present invention is applied for compressing a moving image by MPEG. 1 is an image signal input unit, 2 is a difference unit, 3 is an orthogonal transformer, 4 is a quantizer, 5 is a variable length encoder, 6 is a code amount measuring unit, 7 is a coefficient transforming unit, 8 is a target value setting , 9 is an inverse quantizer, 10 is an orthogonal inverse transformer, 11 is an adder, 12 is a predictor, and 13 is a mode change switch.

There are roughly two types of moving picture coding modes based on MPEG: an intra-frame coding mode and an inter-frame coding mode. In the intra-frame coding mode, the input image is coded as it is, and in the inter-frame coding mode, the already coded image is referred to in one of three prediction modes such as forward, backward, or interpolation of these. A predicted image is generated, and the difference image between this predicted image and the input image is encoded.

In this embodiment, in the intra-frame coding mode, the mode selector switch switch 13 is opened, the difference unit 2 is disabled, and the image data input from the image signal input unit 1 is unchanged as it is in the orthogonal transformer 3. Sent to.
In the inter-frame coding mode, that is, in the mode of performing coding by performing some kind of prediction, the mode selector switch switch 13 is closed, and prediction is performed from a locally decoded image previously stored in a frame memory inside the predictor 12. The difference from the predicted image generated by the differencer 12 is the differencer 2
And the difference image is sent to the orthogonal transformer 3.

The locally-decoded image is an image that is decoded by being locally coded and locally decoded in the image data coding apparatus. This is because when the image data is encoded, the data quantized by the quantizer 4 is converted into the inverse quantizer 9
Is inversely quantized, and the inverse inverse cosine transform is performed by the orthogonal inverse transformer 10, which is stored in the frame memory of the predictor 12 from the already generated locally decoded image to the predictor 1
The predicted image generated in 2 is added to the predicted image by the adder 11 and stored in the frame memory 12.

The DCT coefficient resulting from the discrete cosine transform in the orthogonal transformer 3 is quantized in the quantizer 4,
After variable length coding is performed by the variable length encoder 5, the image data is output to a storage medium or the like. At this time, the code amount measuring device 6
Measures the code amount of variable-length coded data and sends the measurement result to the coefficient conversion unit 7.

The target value setting unit 7 holds a predetermined code amount assigned to each of the picture (I) in the intra-frame coding mode and the picture (B) in the inter-frame coding mode. In the coefficient conversion unit 7, the code amount of each picture sent from the code amount measuring device 6 is set to the target value setting unit 8
It compares with the predetermined code amount held in. The coefficient conversion unit 7 converts the code amount so that if the measured code amount is lower, the code amount is increased so as to increase, and if the measured code amount is higher, the code amount is corrected downward.

Generally, a large amount of code is assigned to a picture (I) in the intra-frame coding mode, which serves as a reference for the quality of reproduced images. The target code amount set in the target value setting unit 8 at this time is as shown in FIG. If the code amount of the input image data is controlled and encoded along the line of the graph of FIG. 4, the entire image data can be controlled to the target code amount.

For example, when the input image is divided into blocks of 8 × 8 pixels and each block is subjected to discrete cosine transform processing by an orthogonal transformer, and the DCT coefficient of the current pixel block is as shown in FIG. think of. The coefficient at the upper left corner of the coefficient matrix at this time is a DC coefficient, and the others are AC coefficients.

In the quantizer 4, the DCT coefficient is
Quantization is performed using a quantization table. However, as a feature of DCT, since the ratio of the amount of image information included in the high frequency component is relatively small, even if the DCT coefficient of the high frequency component is cut off, Since the effect on the visual sense of is smaller than that of the low-frequency component, the rate of image deterioration is relatively small, and especially in the case of moving image data composed of multiple frames, the effect on the overall image quality is suppressed. It

In the code amount control according to the present invention, the coefficient transforming unit 7 truncates the DCT coefficient by converting the high-frequency component of the DCT coefficient into 0 in correspondence with the zigzag scan order.

In this code amount control, the code amount of the current pixel block is measured by the code amount measuring device 6, and based on the measurement result, the high frequency component of the coefficient matrix of the next pixel block, for example, the shaded portion in FIG. The corresponding coefficient value is converted into 0 to generate a coefficient matrix as shown in FIG. The coefficient matrix thus transformed is rearranged in the zigzag scan order in the variable length encoder 5 and zigzag scanned in the order shown in FIG. 5 to perform variable length coding such as Huffman coding.

The code amount measuring unit 6 measures the generated code amount in the variable length encoder 5, and if the code amount is smaller than the information amount designated in advance, controls the coefficient conversion unit 7 to The amount of cut-off DCT coefficients from the pixel block is reduced and the amount of image information is increased. For example,
If the cutoff range of the DCT coefficient of the current pixel block is the shaded area in FIG. 3, the cutoff range from the next block is changed to the shaded area in FIG.

When the code amount is larger than the planned information amount, the coefficient conversion unit 7 is controlled to increase the cutoff amount of the DCT coefficient from the next pixel block and reduce the information amount of the image information. To do so. For example, if the cut-off range of the DCT coefficient of the current block is the shaded area in FIG. 3, the cut-off range of the DCT coefficient from the next block is changed to the shaded area of FIG.

In this way, in the coefficient conversion unit 7, according to the difference from the target code amount set in the target value setting unit 8,
In the order of zigzag scanning, from the highest frequency component,
By converting and truncating the DCT coefficient, it is possible to control the code amount of the image data while minimizing the influence on the image quality.

In the setting of the cut-off range of the DCT coefficient, the number of cut-off coefficients may be increased or decreased by one pixel by pixel, or may be increased or decreased by a predetermined number.

Next, a second embodiment of the present invention will be described. In the present embodiment, the DCT coefficients are divided into 6 groups of a, b, c, d, e, f, and g as shown in FIG. It is done in units.

For example, in the current block, group a
When the code amount is lower than the target code amount when the group b is in the cut-off range and the code amount is higher than the target code amount, the cut-off range from the next block is only the group a. Is controlled so that the cut-off range of the DCT coefficient from the next adjacent block is group a, group b, and group c.

At this time, the target code amount set in the target value setting section 7 and the setting of the cutoff range of the DCT coefficient are set by the coding mode according to the state of the mode changeover switch 13.
It is also possible to separately perform the intra-frame coding picture (I) and the inter-frame coding picture (B).

In the above-mentioned embodiment, the DCT coefficient is cut off in pixel block units. However, when the code amount is prioritized over the image quality, the unit of code amount control is a plurality of blocks in the image of one frame. It is also possible to take it for each block line constituted by. In this case, the code amount measuring device 6 measures the code amount for each block line and controls the DCT coefficient cutoff range for each block line.

Although the moving image processing apparatus has been described in the embodiment of the present invention, m × m such as JPEG.
If it is a method of encoding in pixel block units, it can be easily applied to a still image processing apparatus.

[0034]

As described above, according to the invention described in claim 1, it is possible to control the code amount after compression to the target code amount while suppressing the deterioration of the image quality. Further, since it is not necessary to perform special processing such as fetching the quantization table at the time of decoding, it is possible to decode the image data by the image data decoding device based on the image compression standard such as MPEG and JPEG. , Data compatibility and system extensibility are not impaired.

Further, according to the invention of claim 2, DC
To group the T coefficients and control the truncation range,
The code amount can be controlled more easily than the control at the pixel level.

Further, according to the third aspect of the present invention, the influence of image quality deterioration is caused by changing the cut-off range of the DCT coefficient separately for the picture for intra-frame coding and the picture for inter-frame coding. It is possible to control the code amount so as to approach the target value while suppressing.

[Brief description of drawings]

FIG. 1 is a block configuration diagram of a first embodiment of the present invention.

FIG. 2 is a diagram illustrating a target code amount.

FIG. 3 is an example of a DCT coefficient.

FIG. 4 is an example of a cutoff range of DCT coefficients.

FIG. 5 is a diagram showing the order of zigzag scanning.

FIG. 6 is an example of changing the cutoff range of DCT coefficients.

FIG. 7 is a second modification of the cutoff range of DCT coefficients.

FIG. 8 is an example of grouping DCT coefficients.

[Explanation of symbols]

 1 Image Signal Input Unit 2 Difference Unit 3 Orthogonal Transformer 4 Quantizer 5 Variable Length Encoder 6 Code Amount Measuring Unit 7 Target Value Setting Unit 8 Coefficient Transforming Unit 9 Inverse Quantizer 10 Orthogonal Inverse Transformer 11 Adder 12 Predictor 13 Mode selector switch

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Office reference number FI technical display location H04N 1/41 B 9070-5C

Claims (3)

[Claims] 1. Image data continuously input is m × m.
In an image data processing device that divides into pixel blocks of pixels and encodes image data in pixel block units, code amount measuring means for measuring the code amount generated for each pixel block, and measurement of the code amount measuring means. An image data encoding device characterized by a control means for controlling a cut-off range of image information based on a result. 2. The control means divides the image information of each block into a group consisting of pixel information of a plurality of pixels,
The image data encoding device according to claim 1, wherein the cut-off range of the image information is controlled in units of the groups. 3. The image data according to claim 1, wherein the control means changes a cut-off range of image information between a picture for intra-frame coding and a picture for inter-frame coding. Encoding device.