KR0156137B1 - Quantum coefficient selecting apparatus and method of dvcr - Google Patents

Abstract

The present invention relates to DVCR, and in particular, an error generated during quantization and a coded bit length can be used to efficiently designate an appropriate quantization coefficient (Q _NO ) that minimizes quantization error (Q _E ). the present invention relates to the quantization coefficient selection apparatus and method for a DVCR connected to a quantizer having a quantization of each coefficient (Q _NO15 -Q _NO0) in series by taking the serial processing (serial processing) constituting the video segment by summing bit rate and the quantization error It relates to being able to efficiently specify the quantization coefficients Q _NO of each macro block.

Description

Apparatus and Method for Selecting Quantization Coefficient of DVCR

1 is a block diagram showing a bitstream generation process of a general DVCR.

2 is a block diagram of a macroblock according to NTSC or PAL.

3 (a) and 3 (b) are structure tables showing an example of selection of a quantization step.

4 is a structure table showing quantization steps.

5 is a block diagram showing a conventional quantization coefficient selection device.

6 is a block diagram showing a quantization coefficient selection device of the present invention.

7 is a flowchart showing a method of selecting a quantization coefficient according to the present invention.

* Explanation of symbols for main parts of the drawings

30: Quantizer (Q ₁₅ -Q ₀ ) 31: Bit rate calculator and Q _E adder (15-0)

32: bit estimator

The present invention relates to DVCR, and in particular, an error generated during the quantization process and an appropriate quantization coefficient (Q _NO ) at which the quantization error (Q _E ) is minimized by using the length of the coated bit is efficiently A device and method for selecting a quantization coefficient of a DVCR can be specified.

In general, in the case of standard definition (SD) in 1/4 DVCR, the standard is already determined, and a tape format, an encoding / decoding method, and the like are determined.

That is, the implementation algorithm according to the determined shooting must be developed in each maker itself.

Among them, the problem of determining the quantization coefficient (Q _NO ), which has the greatest impact on the performance of SD, has emerged as an important issue in the implementation of DVCR.

Hereinafter, the quantization coefficient designation in general bitstream generation will be described with reference to the accompanying drawings.

1 is a block diagram showing a bitstream generation process of a general DVCR, and FIG. 2 is a block diagram of a macroblock according to NTSC or PAL.

SD method receives NTSC or PAL signal and creates YUV composite digital signal sampled 4: 2: 2 through A / D conversion, and the sampled image data is reduced to 4: 1: 1 format again. Four 8 × 8 luminance DCT blocks and two 8 × 8 chroma DCT blocks are combined into units of one macro block.

FIG. 2 (a) shows the structure of the macroblock in the NTSC scheme and FIG. 2 (b) shows the structure of the macroblock in the PAL scheme.

The divided image is composed of five macroblocks in a random order to form a video segment.

In this manner, the five macroblocks are combined to form a video segment, which is called shuffling.

The video signals of video segments mixed in macroblock units through the shuffling process as described above are transferred from respective parts of the video segments, and thus have little correlation between video segments. Can be compressed.

The video signal which has been processed in the shuffling section 1 is converted from the spatial domain component to the frequency domain component by the DCT section 2.

At this time, the DCT (Discret Cosine Transform) unit 2 performs a DCT made of a frame (8 × 8) by adding the Even and Odd fields to a portion having less movement according to the characteristics of the input image. Since DCT with separate fields is performed, more effective DCT can be performed.

The video signal output by the DCT conversion of the DCT unit 2 includes a bit rate pre-analyzer 3 that calculates the bit rate of the DCT block and calculates the activity of each block. Multi-Quantizer (5) which allocates substantial bits for efficient compression of the video signal through a normalizer 4 and determines an efficient Q _NO for each macroblock according to the allocated number of bits. Then, the process of performing the bit estimator (Bit Estimator) (6).

Each DCT block that has been transformed by DCT is divided into classes according to its characteristics, and thus, a quantization table according to the class and area for each DCT block, as shown in FIG. ), We select different quantization steps (Quantizer s-tep).

At this time, it must compress the video segment with a predetermined length, and assigns an appropriate bit by five each macroblock constituting the video segment, and determines the Q _NO for each macroblock according to the number of allocated bits, determined by using a Q _NO The DCT blocks are quantized by a quantization converter (Quantizer) 8, and formatted through a Zig-Zag Scan (9) and a Variable Length Coding (10) which performs variable length coding. Formatting is performed in section 11.

4 is a structure table showing the quantization step.

Since the five macroblocks constituting the video segment are shuffled in different parts of the image (frame unit) in the above process, there is no correlation between them, and the video segment is compressed in a certain unit. To do this, the statistical characteristics of the five shuffled macroblocks must be obtained, and a Q _NO that can be coded with a predetermined bit or less for each macroblock must be specified.

That is, as shown in FIG. 5, which is a block diagram showing a conventional quantization coefficient selector, video segments are input to each quantizer (Q _0- Q ₁₅ ) 20, and Class No, Area No is input to each quantizer. Is entered.

That is, it is a multi-quantization system (Multi-Quantizer System) in which all quantization according to Q _NO exists.

The quantizer (Q _0- Q ₁₅ ) 20 and the bit rate calculator (0-15) 21 output quantization steps in each Class No and Area No according to Q _NO to approximate the intended compression ratio. Q _NO is selected by the bit estimator 22.

However, in the conventional multi-quantization system as described above, it is required to shift from 1 bit to a maximum of 4 bit according to the quantization step.

In addition, a certain number of bits must be allocated for each macroblock of an input video segment, and a Q _{NO corresponding} to the number of allocated bits must be selected for each macroblock. If the allocated number of bits does not fit the macroblock, There was a problem that caused distortion in the macro block.

The present invention has been made to solve the problems of the conventional DVCR quantization coefficient selection apparatus and method, using a serial quantizer that does not require a full spec quantization bank (Bank) By using the quantization error (Q _E ) and the coded bit length (Bit Length), the quantization coefficient (Q _NO ) having an optimal quantization error (Q _E ) and having a constant bit rate (Bit Ra-te) can be efficiently An object of the present invention is to provide an apparatus and method for selecting a quantization coefficient of a DVCR.

DVCR quantization coefficient selector of the present invention for achieving the above object has a quantization coefficient (Q _NO15 -Q _NO0 ) is connected in series with a video segment composed of a plurality of macro blocks as input area No and A plurality of quantizers Q ₁₅ -Q ₀ which sequentially perform quantization according to Class No, and a delayer Z ₁₄ -Z ₀ which delays respective outputs of the quantizers Q ₁₅ -Q ₀ for a predetermined time _. ), A plurality of bit rate calculators for calculating a bit rate by the delayer output, and a quantization error by each output of the quantizers Q ₁₅ -Q ₀ connected in series. The quantization system is quantized by the output of the plurality of quantization error adder (15-0) and the bit rate calculator and the quantization error adder are sequentially quantized below the bit rate to be compressed. Characterized by the bit estimation configured to include a selecting a (Q _NO), and the quantized coefficient selection method of the DVCR is in a first step and, for each macroblock of the quantization of the quantization for each macro block constituting the video segment are sequentially in accordance with the bit rate and based on the quantization error (Q _E) a second step, the calculated summed bit rate and the quantization error (Q _E) for calculating a searching the macro block having a minimum quantization error, the quantized coefficient having the appropriate bit rate (quantization Number The third step of specifying a) is characterized in that it is made.

Hereinafter, an apparatus and method for selecting a quantization coefficient of a DVCR according to the present invention will be described in detail with reference to the accompanying drawings.

5 is a block diagram showing a quantization coefficient selection apparatus of the present invention, Figure 7 is a flowchart showing a quantization coefficient selection method according to the present invention.

The apparatus and method for selecting a quantization coefficient of a DVCR according to the present invention is to select a quantization coefficient by using a quantization error and a coded bit length.

As shown in FIG. 4, which is a structure table showing the quantization step, the quantization step exists as 2 ⁿ (n = 0,1,2…) as in 1.2.4.8.16, thus shifting the least significant bit (LSB). quantization process by shifting).

In this case, when the shift circuit LSB is 1, a quantization error is inserted, and when it is 0, a special case without quantization error is provided.

In other words, if the quantization step in the case of 8 (= 1000) is 1.2.4.8, no quantization error occurs.

Therefore, by checking the LSB for each macroblock as described above, the quantization coefficient that minimizes quantization error can be found, so that the quantization coefficient Q _NO of each macroblock that can compress the video segment to a predetermined unit or less can be selected.

Quantization coefficient is the quantization step according to the Q _NO is to be determined by the Class No No Area and inputted when Q _NO15 -Q _NO0 exists.

class No represents the activity of the DCT block, and Area No represents the importance of the DCT coefficients.

Therefore, as shown in FIG. 6, which is a block diagram illustrating a quantization coefficient selection device of the present invention, serial processing is performed by connecting quantizers 30 having respective quantization coefficients Q _NO15 -Q _NO0 in series. Taken, there is always only 1 bit shift, resulting in a simpler hardware of the quantizer 30.

For example, if the quantization coefficient (Q _NO ) is changed from 8 to 7, Class No. The step changes from 1 to 2 only when Area No is 0 to 2. 2, Area No. Since the quantization step is changed from 2 to 4 and 4 to 8 in 1 and 3, only 1 bit shift is performed through each delay Z ₁₄ -Z ₀ to complete the quantization.

And if the rate of five compressed macroblocks is long, i.e., if the compression rate of the video segment is longer than the intended target rate, which macroblock of the five macroblocks is to be compressed (compresses more macroblocks). Means lowering the quantization coefficient (Q _NO )). The quantization error (Q _E) using the LSB calculated by the bit rate calculation / Q _E summing unit 31 during the quantization process of each macro block is described above. ), The bit estimator 32 lowers the Q _NO of the MB having the smallest macroblock quantization error, thereby controlling the bit rate.

The quantization coefficient selection device of the present invention as described above is executed as follows to obtain an optimal quantization coefficient.

As shown in FIG. 7, which is a flowchart illustrating a method of selecting a quantization coefficient according to the present invention, 5 macroblocks are respectively quantized to Q _NO = 15 by receiving an input video segment signal (701). The bit rate and quantization error of MB) are calculated (702).

The sum of the bit rates and the target bit rates is then performed (703).

When the length of compressing the video segment in step 703 is higher than the intended compression rate, that is, when the rate is large, the selection of which MB of the 5 MBs to lower the Q _NO is determined based on the quantization error obtained by the quantization error adder. As a reference, the MB having the minimum Q _E is searched for (705).

And the MB is found with the lowest of the Q _E from the back to the MB having a minimum of _E Q Q _NO is judged whether or not zero (706) Q 0 Q _{NO NO} is a non-zero MBs 707.

The Q _NO of the MB having the minimum Q _E is lowered (708), and the bit rate and Q _E of each macro block are calculated (709), and the summed bit rate and the target bit rate are compared (703). The Q _NO of each macro block is used as the actual scale factor (704).

That is, since Q _NO of the macroblock with the lowest Q _E is lowered, when the compression rate of the video segment is finally adjusted to the target rate, an appropriate quantization coefficient (Q _NO ) that efficiently minimizes the quantization error (Q _E ) can be efficiently specified. It becomes possible.

Claims (3)

A plurality of quantizers (Q _NO15 -Q _NO0 ) which are connected in series with each quantum scribing number (Q _NO15 -Q _NO0 ) and sequentially perform quantization according to each Area No and Class No by inputting a video segment composed of a plurality of macro blocks (Q) ₁₅ -Q ₀ ), a delay (Z ₁₄ -Z ₀ ) for delaying the respective outputs of the quantizers Q ₁₅ -Q ₀ for a predetermined time, and a plurality of bits for calculating the bit rate by the output of the delayer. A plurality of quantization error adders 15-0 connected in series with each of the bit rate calculators to sequentially calculate and quantize quantization errors by respective outputs of the quantizers Q ₁₅ -Q ₀ . And a bit estimator for selecting a quantization coefficient Q _NO that is quantized to be less than or equal to a bit length to compress each macroblock by respective outputs of the bit rate calculator and the quantization error adder. Ugh Quantized coefficient selection apparatus for a DVCR. A first step of sequentially quantizing each macro block constituting the video segment, a second step of calculating a bit rate and a quantization error Q _E according to each of the quantized macro blocks, and a calculated sum of the bit rate and the quantization error Q A method for selecting a quantization coefficient of a DVCR, comprising the step of finding a macroblock having a minimum quantization error based on _E ) and specifying a quantization coefficient to compress each macroblock. . 3. The method of claim 2, wherein the third step includes comparing the sum of the sum bit rate and the destination bit rate to find a macroblock having a minimum quantization error if the sum bit rate is less than the destination bit rate, A fifth step of calculating the bit rate and quantization error of each macroblock by lowering the quantization coefficient of the macroblock having quantization errors; And a seventh step of repeating the sixth step and the fourth to sixth steps, using a quantization coefficient of each macroblock obtained when the sum of the bit rates is smaller than the target bit rate. Quantization Coefficient Selection Method.