CN102256126A

CN102256126A - Method for coding mixed image

Info

Publication number: CN102256126A
Application number: CN2011102001787A
Authority: CN
Inventors: 丁文鹏; 朱维佳; 施云慧; 尹宝才
Original assignee: Beijing University of Technology
Current assignee: Beijing University of Technology
Priority date: 2011-07-14
Filing date: 2011-07-14
Publication date: 2011-11-23

Abstract

The invention discloses a method for coding a mixed image with higher coding efficiency. In the method, a residual scalar quantization (RSQ) mode is used in the interframe coding of a mixed image, according to a principle that rate distortion cost is minimum, each transformation unit (TU) adaptively selects a discrete cosine transform (DCT) mode or the RSQ mode, and a bit is used for expressing the selection; and residual errors obtained after interframe prediction are directly quantized in the RSQ mode, and a step of transforming codes is omitted.

Description

The coding method of vision-mix

Technical field

The invention belongs to the technical field of image processing, relate to a kind of coding method of vision-mix particularly.

Background technology

In existing technology, the coding of vision-mix roughly can be divided into following three classes.

1. based on the method for image encoding:

This method still adopts traditional coding framework to come vision-mix is encoded, but adopt different quantization steps at different zones, for the piece that comprises literal and chart content, the coding back rebuilds effect can be very fuzzy, so reduce quantization step when quantizing these pieces, be its overabsorption bit, make the quality of its reconstruction as well as possible.But for fixing bit number, if the bit that has been literal and the overabsorption of figure table block, the bit number that distributed of natural image piece will correspondingly reduce so, so on the whole, this method still can not reach good effect.

2. based on the method for hierarchical coding:

This method adopts the mixed light bar phantom to compress.It is divided into piece image or image block the plane layer of foreground layer, background layer and a binary system operator.The plane layer of this binary system operator shows that each pixel of image or image block belongs to foreground layer or background layer, and this binary system operator plane can be compressed with JBIG standard or JBIG2 standard.And the foreground layer and the background layer that pass through after the data filling algorithm blurs can compress with traditional image encoding framework.Paper " Compressing compound documents " (The Document and Image Compression Handbook, M.Barni, Ed.New York:Marcel-Dekker, 2005.) this method proposed, and prove that by experiment this method compares with traditional encoding and decoding framework, can obtain very big gain.But this method has some shortcomings: at first gain depends on promptly how image is divided into prospect and background by the algorithm of cutting apart to a great extent.The second, if the effect of coding can be damaged in the hole that splits without any processing.The 3rd, the textcolor of the background layer of encoding respectively and the text shape of binary system operator also can reduce the efficient of coding.So this method still can not satisfy demands of applications well

3. block-based coding method:

This method can be divided into two classes:

A kind of is the character of utilizing the spatial domain, as statistics with histogram, and gradient, the numbers of color etc. can be divided into the present encoding piece a kind of in several classes, as text, chart, natural image etc.Different then classifications is compressed to be applicable to their statistical property, as shown in Figure 1 with different coding frameworks.

Another kind is to consider that the histogram of literal in the vision-mix and chart is more sparse, is the form of base color and concordance list with it in Time Domain Decomposition, i.e. Base color and Index map (BCIM) pattern.

Although block-based coding method has obtained good gain than above-mentioned two kinds of technology, but all this coding methods all have a common ground at present, it is exactly the correlation of only having considered the space, promptly only part in the frame is improved, and ignored the correlation of time, the characteristics of interframe encode are not taken into account, do not designed encoding scheme at the vision-mix of interframe encode.

Summary of the invention

Technology of the present invention is dealt with problems and is: overcome the deficiencies in the prior art, the coding method of the higher vision-mix of a kind of code efficiency is provided.

Technical solution of the present invention is: the coding method of this vision-mix, with RSQ (Residual scalar quantization, the residual error scalar quantization) pattern is used for the interframe encode of vision-mix: according to the principle of rate distortion costs minimum, each TU (Transform unit, converter unit) selects DCT (Discrete cosine transform adaptively, discrete cosine transform) pattern or RSQ pattern, and represent this selection with a bit; The RSQ pattern is that the residual error that will obtain behind the inter prediction quantizes.

Because the invention solves the problem that vision-mix is not suitable for transition coding, and because can be littler through resulting residual error behind the inter prediction, so can be higher for the interframe encode efficient of vision-mix.

Description of drawings

Fig. 1 shows existing RSQ pattern, promptly utilizes the block-based coding method of spatial domain character;

Fig. 2 shows the flow chart according to the coding method of vision-mix of the present invention.

Embodiment

The coding method of this vision-mix, with RSQ (Residual scalar quantization, the residual error scalar quantization) pattern is used for the interframe encode of vision-mix: according to the principle of rate distortion costs minimum, each TU (Transform unit, converter unit) selects DCT (Discrete cosine transform adaptively, discrete cosine transform) pattern or RSQ pattern, and represent this selection with a bit; The RSQ pattern is that the residual error that will obtain behind the inter prediction directly quantizes, and skips the step of transition coding.

When transmission RSQ coding, adopt the row sequential scanning, and distribute context for it.

As shown in Figure 2, the coding method of this vision-mix may further comprise the steps:

(1) for the luminance component among the PU, obtain its prediction piece, and then subtract each other with original value and to obtain residual error, residual block is carried out the division of different size, obtain TU (Transform Unit, converter unit);

(2) residual error among the current TU is quantized;

(3) TU after will quantizing is sent to decoding end;

(4) each TU of decoding end carries out inverse quantization;

(5) TU is combined as PU, carries out addition with the predicted value of the current PU that obtains that passes through motion compensation, obtains the reconstruction value of PU, and then rebuilds current CU (Coding Unit, coding unit).

Preferably, quantification employing formula: C (S)=sign (the S) * max in the step (2) (0, floor (| S|/Q+1-Z), wherein S is an input signal, and floor is downward bracket function, and Q is a quantization step, and Z is the adjusting parameter of presetting.

Preferably, what the transmission in the step (3) was different with the HM standard is without zig-zag sequential scanning residual error, but scans each residual error coefficient successively by raster scan order, then transmission successively.

Preferably, the inverse quantization in the step (4) adopt formula: R (x)=sign (x) * round (q * | x|+p), wherein round is the round function, and q, p are the Control Parameter of the conversion along with quantization parameter.

The above; it only is preferred embodiment of the present invention; be not that the present invention is done any pro forma restriction, every foundation technical spirit of the present invention all still belongs to the protection range of technical solution of the present invention to any simple modification, equivalent variations and modification that above embodiment did.

Claims

1. the coding method of vision-mix, it is characterized in that: the interframe encode that residual error scalar quantization RSQ pattern is used for vision-mix: according to the principle of rate distortion costs minimum, each converter unit TU selects discrete cosine transform DCT pattern or RSQ pattern adaptively, and represents this selection with a bit; The RSQ pattern is that the residual error that will obtain behind the inter prediction quantizes.

2. method according to claim 1 is characterized in that: when transmission RSQ coding, adopt the row sequential scanning, and be its distributing independent context.

3. the coding method of vision-mix is characterized in that: may further comprise the steps:

(1) for the luminance component among the predicting unit PU, obtain its prediction piece, and then subtract each other with original value and to obtain residual error, residual block is carried out the division of different size, obtain converter unit TU;

(2) residual error among the current TU is quantized;

(3) TU after will quantizing is sent to decoding end;

(4) each TU of decoding end carries out inverse quantization;

(5) TU is combined as PU, carries out addition with the predicted value of the current PU that obtains that passes through motion compensation, obtains the reconstruction value of PU, and then rebuilds the coding unit CU of present encoding unit.

4. method according to claim 3 is characterized in that: quantification employing formula: C (S)=sign (the S) * max in the step (2) (0, floor (| S|/Q+1-Z), wherein S is an input signal, floor is downward bracket function, and Q is a quantization step, and Z is the adjusting parameter of presetting.

5. method according to claim 3 is characterized in that: the transmission in the step (3) is to scan each residual error coefficient successively by raster scan order, then transmission successively.

6. method according to claim 4, it is characterized in that: inverse quantization employing formula: R (x)=sign (the x) * round in the step (4) (q * | x|+p), wherein round is the round function, and q, p are the Control Parameter of the conversion along with quantization parameter.