WO2016000210A1 - An improved depth lookup table - Google Patents

Abstract

It is proposed to clip the reconstructed index of the depth lookup table (DLT) to a proper value. In this case, meaningless DLT indices can be avoided.

Description

AN IMPROVED DEPTH LOOKUP TABLE

FIELD OF INVENTION

The invention relates generally to Three-Dimensional (3D) video processing. In particular, the presented invention relates to depth coding.

BACKGROUND OF THE INVENTION

Depth lookup table (DLT) is adopted into 3D-HEVC. Since there are often only several values appearing in the depth component, DLT signals those valid values from the encoder to the decoder. When a CU is coded in intra simplified depth coding (SDC) mode or depth map modeling (DMM) mode, DLT is used to map the valid depth value to a DLT index which is much easier to compress. Fig.1 demonstrates an example of DLT approach. DLT is signaled in picture parameter set (PPS). And it is left to be an encoder issue how to get the DLT when encoding.

In the current design, the prediction samples are calculated as follows:

predSamples[ x ][ y ]= Idx2DepthValue[ Depth Value2Idx[ predDcVal ] + dcOffset ], wherein (x, y) represents the sample coordination, predDcVal is the DC prediction value, dcOffset is an offset signaled from the encoder to the decoder, Depth Value2Idx is a mapping table that maps a depth value to a DLT index, and Idx2Depth Value is a mapping table that maps a DLT index to a depth value.

In such a design, Depth Value2Idx[ predDcVal ] + dcOffset might exceed the range of valid DLT indices. In such a case, Idx2Depth Value [ Depth Value2Idx [ predDcVal ] + dcOffset ] will try to map an invalid DLT index to a depth value. The mapping will fail and the decoding process cannot run correctly. For example, when there are only 10 entries in DLT, the valid DLT indices are 0, 1, ...8, 9. If Depth Value2Idx [ predDcVal ] + dcOffset ] is equal to 20 or -1, the value of Idx2DepthValue[ 20 ] is undefined and the decoding process will be crashed.

JCT3 V-I0057 was proposed to address this problem. In JCT3 V-I0057,

predSamples[ x ][ y ]= Idx2DepthValue[ Depth Value2Idx[ predDcVal ] + dcOffset ] is replaced by

predSamples[ x ][ y ]=Idx2DepthValue[ Clip3( 0, ( 1 « bitDepth ) - 1, Depth Value2Idx[ predDcVal ] + dcOffset ) ],

wherein bitDepth is the bit depth of depth samples. For example, bitDepth is 8 when depth values are from 0 to 255.

However, JCT3V-I0057 does not solve the problem completely. It make a constrain that Depth Value2Idx[ predDcVal ] + dcOffset cannot exceed ( 1 « bitDepth ) - 1. But the number of DLT entries usually is smaller than ( 1 « bitDepth ). For example, when bitDepth is 8 then the prediction samples are calculated in JCT3V-I0057 as

predSamples[ x ][ y ]= Idx2DepthValue[ Clip3( 0, 255,DepthValue2Idx[ predDcVal ] + dcOffset ) ].

When there are only 10 entries in DLT, the valid DLT indices are 0, 1, ...8, 9. If Depth Value2Idx [ predDcVal ] + dcOffset ] is equal to 20, the value of Idx2DepthValue[ 20 ] is still undefined and the decoding process will be crashed.

SUMMARY OF THE INVENTION

In light of the previously described problems, methods are proposed to improve the depth lookup table (DLT).

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 THE DRAWINGS

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

Fig. 1 is a diagram illustrating an example of DLT. There are five valid values appears in depth samples Therefore, there are five entries in DLT in the order: 50, 108, 110, 112 and 200;

Fig. 2 is a diagram illustrating the flowchart to clip the DLT index to a valid value;

Fig. 3 is a diagram illustrating an example of the original Idx2Depth Value table and the proposed Idx2Depth Value table for i >= NumDepthValuesInDlt. NumDepthValuesInDlt=5 in the example;

Fig. 4 is a diagram illustrating an example of the original Idx2Depth Value table and the proposed Idx2Depth Value table for i < 0. NumDepthValuesInDlt=5 in the example. DETAILED DESCRIPTION

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.

It is proposed to make the prediction samples always valid when depth lookup table is applied.

In one embodiment, the derivation of prediction samples as

predSamples[ x ][ y ]= Idx2DepthValue[ Depth Value2Idx[ predDcVal ] + dcOffset ] is replaced by

predSamples[ x ][ y ]=Idx2DepthValue[Clip3(0, NumDepthValuesInDlt[ nuh layer id ]-l, DepthValue2Idx[ predDcVal ] + dcOffset ) ],

wherein NumDepthValuesInDlt[ nuh layer id ] is the number of depth values in the DLT for the current coding layer nuh layer id. Fig.2 illustrates the clipping process.

In another embodiment, the DC values for two segments in segment-wise depth coding (SDC) or depth-map model (DMM) are calculated as

DcVal[ 0 ] = Idx2DepthValue[ Clip3( 0, NumDepthValuesInDlt[ nuh layer id ] - 1, DepthValue2Idx[ dcValLT ] + ( dcOffset AvailFlag ? DcOffset[ xTb ][ yTb ][ 0 ] : 0 ) ) ];

DcVal[ 1 ] = Idx2DepthValue[ Clip3( 0, NumDepthValuesInDlt[ nuh layer id ] - 1,

DepthValue2Idx[ dcValBR ] +( dcOffsetAvailFlag ? DcOffset[ xTb ][ yTb ][ 1 ] : 0 ) ) ].

In still another embodiment, Idx2DepthValue[ i ] is set equal to DltDepthValue[ NumDepthValuesInDlt- 1] for all i larger than or equal to NumDepthValuesInDlt as illustrated in Fig. 3. Idx2Depth Value is the table mapping a DLT index to a depth value. NumDepthValuesInDlt is the number of entries in the current DLT. DltDepth Value is the table mapping a depth value to a DLT index.

In still another embodiment, the list elements Idx2DepthValue[ i ] specifying the depth value of the i-th index in the lookup table with i ranging from NumDepthValuesInDlt[ nuh layer id ] to ( 1 « bitDepth ) - l, inclusively, is set equal to DltDepth Value[ nuh layer id ][ NumDepthValuesInDlt[ nuh layer id ]- 1].

In still another embodiment, Idx2DepthValue[ i ] is set equal to DltDepth Value[ 0 ] for all i lower than 0 as illustrated in Fig. 4. NumDepthValuesInDlt. Idx2Depth Value is the table mapping a DLT index to a depth value. DltDepth Value is the table mapping a depth value to a DLT index.

In still another embodiment, the list elements Idx2DepthValue[ i ] specifying the depth value of the i-th index in the lookup table with i ranging from l-( 1 « bitDepth ) to - 1, inclusive, is set equal to DltDepth Value[ nuh layer id ][0].

Claims

1. A method of depth coding, comprising mapping a valid depth value from a reconstructed depth lookup table (DLT) index.

2. The method as claimed in claim 1, wherein the reconstructed DLT index is clipped to a valid DLT index before it is used to map a depth value.

3. The method as claimed in claim 2, wherein the reconstructed DLT index is clipped to NumDepthValuesInDlt - 1 if it is larger than or equal to NumDepthValuesInDlt, where NumDepthValuesInDlt is a number of entries in the current DLT.

4. The method as claimed in claim 1, wherein Idx2DepthValue[ i ] is set equal to DltDepth Value[ NumDepthValuesInDlt- 1] for all i larger than or equal to NumDepthValuesInDlt; Idx2Depth Value is the table mapping a DLT index to a depth value; NumDepthValuesInDlt is a number of entries in the current DLT; and DltDepth Value is the table mapping a depth value to a DLT index.

5. The method as claimed in claim 1, wherein Idx2DepthValue[ i ] is set equal to DltDepth Value[ 0 ] for all i lower than 0; NumDepthValuesInDlt; Idx2Depth Value is the table mapping a DLT index to a depth value; and DltDepth Value is the table mapping a depth value to a DLT index.