WO2014205665A1

WO2014205665A1 - An efficient coding method for dlt in 3dvc

Info

Publication number: WO2014205665A1
Application number: PCT/CN2013/077983
Authority: WO
Inventors: Kai Zhang; Jicheng An
Original assignee: Mediatek Singapore Pte. Ltd.
Priority date: 2013-06-26
Filing date: 2013-06-26
Publication date: 2014-12-31

Abstract

This contribution presents a new coding method named enhanced range constrained bit-map (ERCBM) to code DLT more efficiently. In ERCBM, DLT is represented by a range constrained bit-map (RCBM). The bit-map can be coded more efficiently by applying differential run-length coding. In addition, delta-DLT approach can be applied on the dependent view, in order to utilize the inter-view correlation without introducing any potential parsing dependency. Experimental results show that the proposed method provides the similar coding performance to the anchor. In another comparison, the average number of bits in DLT is reduced from 557 to 50.

Description

AN EFFICIENT CODING METHOD FOR DLT IN 3DVC

TECHNICAL FIELD

[0001] The invention relates generally to Three-Dimensional (3D) video processing. In particular, the present invention relates to improvement on Depth Lookup Table (DLT) coding.

BACKGROUND

[0002] Simplified Depth Coding (SDC) and a Depth Lookup Table (DLT) are adopted into HTM. For each CU, if SDC mode is selected, there is only a set of four different prediction modes available. After having selected the optimal prediction modes among the four, the resulting residual is not coded using transform. As the SDC prediction stage always results in one or two depth segments per coded block, a single residual DC depth value is coded for each of these segments. Moreover, DLT is used to map coded depth values in SDC to valid depth values of the original depth map. The DLT is constructed based on an initial analysis of the input depth map and is then coded in SPS.

[0003] DLT is an optional coding tool. In the current HTM, the encoder will not use DLT if all values from 0 to MAX_DEPTH_VALUE (e.g. 255) appear in original depth map at the analysis step. Otherwise, DLT will be coded in SPS. Figure 1 illustrates an example of DLT. In order to code DLT, the number of valid depth values is coded with Exp-Golomb code first. Then each valid depth value is coded also with Exp-Golomb code as depicted in Table 1.

[0004] Exp-Golomb code is efficient when smaller values possess higher probabilities of occurrence. However, that is not the case for valid depth values and the number of valid depth values. As a result, Exp-Golomb codes usually have long code-lengths thus DLT requires a large amount of coding bits. In the common test condition, four sequences Balloons, Kendo, Newspapercc and PoznanHall are determined to use DLT, while the other three are not. Statistics are done on the four sequences. As depicted in Table 2, DLT requires 557.33 bits in average, which account for more than 65% bits of SPS in depth components. In other words, the DLT coding makes SPS in depth components much larger than the one without DLT.

Table 1 DLT coding method in the current HTM.

Table 2 Number of bits in DLT and SPS in the current HTM.

SUMMARY

[0005] In light of the previously described problems, an efficient coding method for DLT in 3DVC is proposed. It is proposed to code DLT using a Enhanced Range Constrained Bit-Map (ERCBM) coding method. The proposed method can reduce the number of bits required by DLT significantly.

[0006] Other aspects and features of the invention will become apparent to those with ordinary skill in the art upon review of the following descriptions of specific embodiments. BRIEF DESCRIPTION OF DRAWINGS

The invention can be more fully understood by reading the subsequent detailed description and examples with references made to the accompanying drawings, wherein:

[0007] Fig. 1 is a diagram illustrating an example of DLT. There are five valid values appears in depth samples at the analysis step. Therefore, there are five items in DLT in the order: 50, 108, 110, 112 and 200.

[0008] Fig. 2 is a diagram illustrating an an example of 0-run-lengths in the bit- map.

[0009] Fig. 3 is a diagram illustrating delta-DLT coding.

DETAILED DESCRIPTION

[0010] The following description is of the best-contemplated mode of carrying out the invention. This description is made for the purpose of illustrating the general principles of the invention and should not be taken in a limiting sense. The scope of the invention is best determined by reference to the appended claims.

[0011] To code the DLT more efficiently, an enhanced range constrained bit-map (ERCBM) coding method is proposed. With ERCBM, DLT is represented by a range constrained bit-map as proposed in JCT3V-C0142 and JCT3V-D0151. min_dlt_value and diff_max_dlt_value are coded as unsigned integer to constrain the value range [MinDltValue, MaxDltValue] of DLT. Values out of this range are not in DLT. Values in DLT between (MinDltValue, MaxDltValue) can be represented by a bit- map BM[i] with from 0 to MaxDltValue - MinDltValue - 2. If BM[i = 1, then value ( + MinDltValue + 1) belongs to DLT. Otherwise, value ( + MinDltValue + 1) does not belong to DLT.

[0012] A differential run-length coding approach is applied to code the bit-map. Obviously, 0-run-lengths can represent the bit-map uniquely. In order to code these 0- run-lengths efficiently, the minimum O-run-length is coded firstly, and then differences between 0-run-lengths and the minimum one are coded sequentially. Fig.2 demonstrates an example of a bit-map, where 0-run-lengths are 3, 2, 2, 3 sequentially. The minimum one '2' is coded firstly, and then the differences 1, 0, 0, 1 are coded in order. The differences are coded by fixed-length coding with n bits. If a difference exceeds 2"-2, the excess will be coded with Exp-Golomb code, n is signaled before differences are coded. Thus the encoder can choose the optimal n which generates the minimum number of bits.

[0013] To guarantee a predictable as well as acceptable number of bits in the worst case, the encoder can choose to code the RCBM directly. In this way, a DLT is coded with 265 bits in the worst case for 8-bit depth samples. A conformable encoder should never code a DLT with more than 265 bits.

[0014] In addition, the delta-DLT method proposed in JCT3V-D0054 is adopted on the dependent view to utilize the inter-view correlation. As depicted in Fig. 3, DLT0 and DLT1 are DLTs on the base view and on the dependent view respectively. They are both represented in a full bit-map form. DLT1' is calculated as DLT1' =

DLT0 ® DLT1. Then DLT1' instead of DLT1 is passed to the DLT coder. After DLT1' is decoded, DLT1 is reconstructed as DLT1 = DLT0 ® DLT1' . Since DLT0 and DLT1 are quite similar, the exclusive or operation will result a lot of zeros, which can benefits the 0-run-length coding. It should be noted that the inter- view information is only utilized at the reconstruction step, so delta-DLT approach does not introduce any parsing dependency issue. Therefore, ERCBM can work well no matter DLTs are coded in VPS, SPS, PPS or slice headers.

[0015] Syntax elements of the proposed method are demonstrated in Table 3.

[0016] The proposed method described above can be used in a video encoder as well as in a video decoder. Embodiments of methods according to the present invention as described above may be implemented in various hardware, software codes, or a combination of both. For example, an embodiment of the present invention can be a circuit integrated into a video compression chip or program codes integrated into video compression software to perform the processing described herein. An embodiment of the present invention may also be program codes to be executed on a Digital Signal Processor (DSP) to perform the processing described herein. The invention may also involve a number of functions to be performed by a computer processor, a digital signal processor, a microprocessor, or field programmable gate array (FPGA). These processors can be configured to perform particular tasks according to the invention, by executing machine-readable software code or firmware code that defines the particular methods embodied by the invention. The software code or firmware codes may be developed in different programming languages and different format or style. The software code may also be compiled for different target platform. However, different code formats, styles and languages of software codes and other means of configuring code to perform the tasks in accordance with the invention will not depart from the spirit and scope of the invention.

[0017] The invention may be embodied in other specific forms without departing from its spirit or essential characteristics. The described examples are to be considered in all respects only as illustrative and not restrictive. To the contrary, it is intended to cover various modifications and similar arrangements (as would be apparent to those skilled in the art). Therefore, the scope of the appended claims should be accorded the broadest interpretation so as to encompass all such modifications and similar arrangements.

Table 3 the proposed DLT coding method.

Claims

1. A method of coding a lookup table which records specific integer values in order, wherein the lookup table is represented and coded as a bit-map and the lookup table is derived by a decoded bit-map.

2. The method as claimed in claim 1, wherein the lookup table is used for depth intra-coding in three dimensional video coding (3DVC), and the lookup table is Depth Lookup Table (DLT) in 3DVC.

3. The method as claimed in claim 1, wherein a difference between a largest value and a smallest value in the lookup table, noted as diff_value is coded.

4. The method as claimed in claim 3, wherein diff_value is coded as an integer or unsigned integer with n bits, where n = |^~log₂ (MAX _ VALUE + 1)^~| .

5. The method as claimed in claim 1, wherein a smallest value in the lookup table noted as min_value is coded.

6. The method as claimed in claim 5, wherein min_value is coded as an integer or unsigned integer with n bits, where n = |^~log₂ (MAX _ VALUE + 1 - diff _ value ) ,

7. The method as claimed in claim 1, wherein a largest value in the lookup table is coded explicitly or inferred implicitly.

8. The method as claimed in claim 3, the largest value in the lookup table noted as MaxV is set equal to MinV+diff_value, where MinV is equal to the smallest value in the lookup table.

9. The method as claimed in claim 1, wherein BitMapFlag[i] with i from 0 to MAX_VALUE is set equal to 0 initially, where BitMapFlag represents a bit-map and MAX_VALUE is a predefined possible largest value in the lookup table, and BitMapFlag[MinV] and BitMapFlag[MaxV] are set equal to 1.

10. The method as claimed in claim 1 wherein a flag ran_length_coding_flag is coded to signal whether run-length coding is used or not.

11. The method as claimed in claim 8, wherein the bit-map in the range (MinV,

MaxV) is coded directly noted as bit_map_flag[j] if run-length coding is not used; BitMapFlag[j] is set equal to bit_map_flag[j] for j from MinV+1 to MaxV-1; if MaxV is greater than MinV, there are MaxV- MinV-1 bit_map_flags coded; otherwise, bit_map_flag is not coded.

12. The method as claimed in claim 8, wherein 0-run-lengths are signaled if run- length coding is used; 0-run-length indicates the number of O's between two Ts in the bit-map; 0-run-length is 0 between two adjacent Ts in the bit-map; and the bitmap in the range (MinV, MaxV) is derived from the coded 0-run-lengths.

13. The method as claimed in claim 12, wherein a minimum 0-run-length is coded; and all 0-run-lengths are greater or equal to the minimum one, noted as MinRun.

14. The method as claimed in claim 12, wherein a difference between each 0- run-length and MinRun is coded.

15. The method as claimed in claim 14, wherein the difference is coded as an unsigned integer by fixed-length coding with N bits, denoted as run_length_diff[j]; if the difference is equal or larger than 2^N-1, a remainder number is coded by Exp- Golomb code, denoted as run_length_diff _rem[j]; Run[j] is set equal to run_length_diff[j]+run_length_diff_rem[j]+MinRun; BitMapFlag[j+Run[j]] is set equal to 1 ; BitMapFlag[i] with i from j to j+Run[j]-l is set equal to 0, if Run[j] is not equal to 0.

16. The method as claimed in claim 15, wherein N is coded before any differences is coded.

17. The method as claimed in claim 1, wherein the bit-map is processed by some operations before it is coded; the decoded bit-map is processed by reverse-operations to reconstruct the identical bit-map on an encoder before processing.

18. The method as claimed in claim 17, the operation and reverse- operation pairs include but not limited to

1 ) Encoder: A'=~A; Decoder: A=~A'.

2) Encoder: A'=A @ B; Decoder: A=A' @ B.

3 ) Encoder: A'=A-B; Decoder: A=A+B. where A represents the bit-map before processing, A' is the bit- map after processing, and B is another bit-map which is obtained both at encoder and decoder, ~ means reversion, and @ means exclusive or.

19. The method as claimed in claim 1, wherein a value equal to i is one item in the lookup table if BitMapFlag[i] =1, and a value equal to i is not one item in the lookup table if BitMapFlag[i] =0.

20. The method as claimed in claim 1, wherein a value equal to i is one item in the lookup table if BitMapFlag[i] =0, and a value equal to i is not one item in the lookup table if BitMapFlag[i] =1.