JPH11298896A - Image processing unit - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce effectively compression distortion in a digital VTR especially in a high compression mode. SOLUTION: The image processing unit is provided with a base band processing section 2 that applies filter processing to an image signal to eliminate its high frequency component, a discrete cosine transform DCT processing section 5, a quantization processing section 6, and a variable length coding VLC processing section 7 that apply compression processing to the image signal after the filter processing, and a microcomputer 9 that controls the filter characteristic based on the compression level in the compression processing. Then a fogged image is reduced by increasing the filter processing amount when the compression level is high to reduce compression distortion and reducing the filter processing amount when the compression level is small.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus applied to, for example, a consumer digital VTR (video tape recorder).

[0002]

2. Description of the Related Art In recent years, consumer digital video tape recorders (so-called DV tape recorders) have become widespread.
A digital VTR) can record and reproduce for a longer time with higher quality than a conventional analog VTR.

In the recording signal processing of the consumer digital VTR, an image signal generated by a camera unit or an externally supplied image signal is processed by baseband processing, blocking processing, shuffling processing, DCT (discrete cosine transform). ) Processing, quantization processing, VLC (variable length coding) processing, etc., and are recorded on a magnetic tape in a predetermined tape format.

On the other hand, an image signal reproduced from a magnetic tape on which the image signal is recorded is subjected to VLD (decoding of a variable length code) process, inverse quantization process, IDCT (inverse discrete cosine transform) process, deshuffling process, After being subjected to a deblocking process or the like, it is sent to, for example, a monitor and displayed.

As described above, at the time of recording a digital VTR, the image signal is compressed by a compression process including DCT processing, quantization processing, and VLC processing, and at the time of reproduction, the compressed image signal is compressed. Is decompressed by a decompression process including a VLD process, an inverse quantization process, and an IDCT process.

[0006]

There are two types of compression in the digital VTR, a standard mode (SD mode) as described later and a high compression mode (SDL mode) having a higher compression ratio than that. I do. In particular, the compression ratio is high in the high compression mode, and therefore, when the image data compressed in the high compression mode is reproduced later, the compression distortion becomes more conspicuous.

As a method for reducing such compression distortion, there is a method for removing high-frequency components of an image signal by, for example, filtering.

However, if the high-frequency components of the image signal are removed by the filter processing, the reproduced image will always look as if it is entirely in love (for example, the outline of the image will not be clear). On the other hand, if the degree of the filtering process is reduced, the image is less in love, but the compression distortion is more conspicuous.

Accordingly, the present invention has been made in view of such circumstances, and it is an object of the present invention to provide an image processing apparatus capable of effectively reducing compression distortion particularly at high compression. .

[0010]

According to the present invention, there is provided an image processing apparatus comprising: a filter processing means for filtering an image signal;
The above-described object is achieved by providing a compression processing unit that compresses an image signal after the filtering process and a filter characteristic control unit that controls a filter characteristic based on a compression level of the compression process.

Here, the filtering process is a process for removing a high-frequency component of the image signal, and the filter characteristic control means increases the removal amount of the high-frequency component when the compression level is larger than a predetermined threshold, and conversely compresses the high-frequency component. When the level is smaller than a predetermined threshold, the removal amount of the high frequency component is reduced.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the present invention will be described with reference to the drawings.

FIG. 1 shows a schematic configuration of a digital VTR as an embodiment to which the image processing apparatus of the present invention is applied. FIG. 1 shows only a main part of a digital VTR according to the image processing apparatus of the present invention, and does not show a configuration for recording and reproducing on a magnetic tape, a reproduction signal processing system, and the like.

In FIG. 1, an image signal supplied from a camera unit (not shown) having an imaging CCD (solid-state imaging device) and a lens system and an external line input unit is supplied to a baseband processing unit 2 via a terminal 1. Sent. The baseband processing unit 2 performs analog / digital conversion processing of sampling the supplied Y signal (luminance signal) and C signal (color difference signal) of the supplied image signal at a predetermined sampling frequency and converting them into digital image data, Perform filter processing and the like. The digital image data from the baseband processing unit 1 is sent to the blocking processing unit 3.

The block processing unit 3 divides the image data supplied from the baseband processing unit 1 into macroblocks, which are basic units of DCT (discrete cosine transform) described later. The image data for each macroblock from the blocking unit 3 is sent to the shuffling unit 4.

The shuffling processing section 4 collects the five macroblocks constituting one video segment by shuffling them from a frame image according to a predetermined rule. The purpose of this shuffling is to equalize the amount of information contained in five macroblocks in a video segment, which is a unit of fixed length. In the shuffling, first, an image for one frame is divided vertically into five macroblocks in a video segment (divided into five parts), and the number of tracks on a magnetic tape for recording one frame of data in the horizontal direction is obtained. (NT
In a 525/60 system such as SC system, 10 divisions,
12 in a 625/50 system such as the PAL system). One block thus divided is called a super block. In shuffling, the first macroblock in each of the five superblocks is collected to form a first video segment, and the next video segment is formed by collecting the next macroblock in the same five superblocks. After collecting all the macro blocks from the five super blocks in this way,
The processing moves to the next five super blocks. The image data shuffled by the shuffling processing section 4 is sent to a DCT (discrete cosine transform) processing section 5.

The DCT processing section 5 performs DCT processing on each macroblock in each video segment (5 macroblocks) formed by the shuffling processing as described later, and obtains a frequency component ( DCT coefficient data) in a so-called zigzag scan order. The DCT coefficient data output from the DCT processing unit 5 and arranged in the zigzag scan order is sent to the quantization processing unit 6.

The quantization processing unit 6 quantizes the DCT coefficient data based on a quantization table to be described later, generates quantized data, and supplies it to a VLC (variable length coding) processing unit 7. Further, the quantization processing unit 6 performs quantization that indicates a class number (class No.) and an area number (area No.) at the time of classification used in a quantization table to be described later, and a selected quantizer. The number (quantization No.) is extracted as compression information indicating the level of compression, and is supplied to the microcomputer 9.

The VLC processing section 7 performs variable length coding on the quantized data supplied from the quantization processing section 6.
The code data obtained by the variable length coding is supplied to a terminal 8
After that, for example, a framing process, a deshuffling process, an error correction encoding process, and the like are performed, and further, a predetermined recording modulation is performed, and the resultant is sent to a recording system.
The data is recorded on the magnetic tape in a predetermined tape format.

Further, the microcomputer 9 generates a filter coefficient for filter processing in the baseband processing unit 2 based on the compression information from the quantization processing unit 6.

That is, the baseband processing unit 2 filters out high-frequency components of an image signal to reduce compression distortion caused by image compression processing in the digital VTR. If you apply too much bandpass processing, the whole image will always fall in love,
On the other hand, if the degree of the filter processing is reduced, the compression distortion becomes more conspicuous. On the other hand, the microcomputer 9 sets a threshold of the compression level (a value corresponding to the compression level that can make the image inadvertent and the compression distortion less noticeable). )
Is calculated, and the threshold value of the compression level and the quantization processing unit 6 are calculated.
The most appropriate filter coefficient is selected based on the compression information from (i.e., the filter characteristics are varied according to the compression level).

That is, the digital VTR of the present embodiment
Then, the compression information in the compression processing is fed back to the microcomputer 9 and the filter coefficient is varied based on the threshold of the compression level and the compression information. For example, when there is a possibility that the compression distortion becomes conspicuous, the filter processing is performed. When the compression distortion is not conspicuous, the amount of filter processing is reduced so that the image quality is reduced.

The operation of the above-described digital VTR will be described in detail with reference to FIGS.

There are two compression modes in the digital VTR, a standard mode (SD mode) as shown in FIG. 2 and a high compression mode (SDL mode) as shown in FIG. Note that FIGS. 2 and 3
Are images of one frame of image data.

In the case of the standard mode shown in FIG.
One frame of image data consists of 10 continuous
In the high compression mode shown in FIG. 3, one frame of image data is divided into five tracks and recorded on a magnetic tape. Will be recorded. That is, from this, in the case of the high compression mode shown in FIG.
It can be seen that the data amount is compressed to half that in the case of the standard mode shown in FIG.

Here, one track is composed of five clusters of 27 macroblocks. Even if the size of one macroblock is the standard mode shown in FIG. 2 or the high compression mode shown in FIG. Is the same size in the case of.

The macro blocks in the standard mode are configured as shown in FIG. 4, and the macro blocks in the high compression mode are configured as shown in FIG. That is, the macroblock in the standard mode shown in FIG. 4 has four samples of a Y signal (luminance signal) of 14 bytes each and a C block of 10 bytes each.
The macroblock in the high compression mode shown in FIG. 5 is composed of 6 samples each of a 10-byte Y signal and 2 samples each of an 8-byte C signal. .

Therefore, in order to convert an image signal in the standard mode into an image signal in the high compression mode, 14 bytes of the Y signal in the standard mode are replaced with 10 bytes of the Y signal in the high compression mode, and the C signal in the standard mode is changed. 10 bytes of C signal of high compression mode
Must be compressed into bytes.

The DCT processing unit 5 shown in FIG.
Each DCT coefficient data in 8 × 8 pixel units in both horizontal and vertical directions obtained by performing DCT processing on a macro block is divided into areas according to frequency components. In the case of the standard mode, as shown in FIG. Each DCT coefficient data is assigned an area number from 0 to 4 (area number).
Are divided into four areas, and in the case of the high compression mode, each DCT coefficient data is numbered 0 to
8 given area number (area No.) up to 7
It is divided into individual areas.

Next, in the case of the standard mode divided into four areas as shown in FIG. 6, the DCT coefficient data in each of these areas is converted by the quantization processing section 6 in FIG. Quantization based on the quantization table. Further, in the case of the high compression mode divided into eight areas as shown in FIG. 7, the DCT coefficient data of each area is obtained by the quantization processing unit 6 of FIG. 1 based on the quantization table shown in FIG. Will be quantized.

The quantization tables shown in FIGS. 8 and 9 are made up of a quantization number (quantization number), a class number (class number), and an area number (area number). The quantization number corresponds to the quantization step,
The quantization step becomes finer as the value of the quantization number increases. As shown in FIGS. 6 and 7, the area number is assigned a shorter code length as the value of the area number is larger, that is, as the DCT coefficient data of a higher frequency component is assigned. That is, the smaller the value of the quantization number and the larger the area number, the higher the compression ratio (the higher the compression level) and the more noticeable the compression distortion.

Therefore, the microcomputer 9 of FIG. 1 changes the filter coefficient used in the baseband processing unit 2 of FIG. 1 while monitoring the compression information of the quantization number and the area number. More specifically, when the quantization number indicated by the compression information is larger than the predetermined number and the area number is smaller than the predetermined number, the microcomputer 9 reduces the filter processing amount because the compression distortion is inconspicuous. On the other hand, when the quantization number indicated by the compression information is smaller than the predetermined number and the area number is larger than the predetermined number, the compression distortion is conspicuous, so that the filter processing amount is increased and the compression is performed. Control to reduce distortion.

In the digital VTR according to the present embodiment, by performing the filter coefficient control based on the above-described compression information, while reducing the compression distortion particularly in the high compression mode, the image is overwhelmed by the over-filtering. Is minimized, and as a result, image quality is improved.

[0034]

As is apparent from the above description, according to the present invention, the image signal after the filter processing is compressed, and the filter characteristics are feedback-controlled based on the compression level of the compression processing. It is possible to effectively reduce the compression distortion at the time of high compression in the digital VTR format.

[Brief description of the drawings]

FIG. 1 is a block circuit diagram showing a schematic configuration of a main part of a digital VTR according to an embodiment of the present invention.

FIG. 2 is a diagram illustrating an image of data for one frame in a standard mode.

FIG. 3 is a diagram showing an image of data for one frame in a high compression mode.

FIG. 4 is a diagram used to describe a macroblock in a standard mode.

FIG. 5 is a diagram used to describe a macroblock in a high compression mode.

FIG. 6 is a diagram used to explain DCT coefficient data and area division in a standard mode.

FIG. 7 is a diagram used to explain DCT coefficient data and area division in a high compression mode.

FIG. 8 is a diagram used for describing a quantization table in the standard mode.

FIG. 9 is a diagram used to describe a quantization table in a high compression mode.

[Description of Signs] 2 Baseband processing unit, 3 Blocking processing unit,
4 Shuffling processing unit, 5 DCT processing unit, 6
Quantization processing unit, 7 VLC processing unit, 9 microcomputer

Claims (4)

[Claims] 1. An image processing apparatus for compressing an image signal, comprising: a filter processing means for filtering the image signal; a compression processing means for compressing the filtered image signal; and a compression level in the compression processing. And a filter characteristic control means for controlling a filter characteristic of the filter processing. 2. The method according to claim 1, wherein the filter processing is processing for removing a high-frequency component of the image signal, and the filter characteristic control means controls the filter characteristic when the compression level is larger than a predetermined threshold value to control the high-frequency component. 2. The image processing apparatus according to claim 1, wherein the removal amount of the high-frequency component is reduced by controlling the filter characteristic when the compression level is smaller than a predetermined threshold. 3. The compression processing includes a standard compression processing and a high compression processing, and the filter characteristic control means controls the filter characteristics during the high compression processing. Image processing device. 4. The compression processing includes a discrete cosine transformation processing and a quantization processing, and the compression processing means determines the frequency area information of the discrete cosine transformation processing and the quantization step of the quantization processing by the compression level. 2. The image processing apparatus according to claim 1, wherein the information is supplied to the filter characteristic control means as information representing