CN102790892A - Depth map coding method and device - Google Patents

Abstract

The present invention proposes a depth map encoding method and device, the method comprising: establishing a plurality of division lines and forming a division set, and the plurality of division lines are used for wedge-shaped division of depth macroblocks; performing intra-coding mode on depth macroblocks Encoding to obtain the first rate-distortion cost value; judge whether to encode the depth macroblock in inter-frame coding mode, and if so, encode in inter-frame coding mode to obtain the second rate-distortion cost value; continue to judge whether the depth macroblock contains discontinuity If yes, encode the depth macroblock in the geometric division coding mode, including: select the optimal division line to divide the depth macroblock to obtain the first and second depth sub-regions, and perform predictive coding on the two sub-regions Obtain a third rate-distortion cost value; compare the rate-distortion cost values in different coding modes to select a coding mode pair with the smallest rate-distortion cost for coding. The embodiments of the present invention improve the compression efficiency of the depth map and reduce the coding complexity.

Description

Depth map coding method and device

technical field

The present invention relates to the technical field of video coding, in particular to a depth map coding method and device.

Background technique

In stereoscopic video and free-viewpoint video systems, the data format composed of multi-viewpoint video and multi-viewpoint depth map has been widely used. Wherein, the depth refers to the distance from each pixel in the video frame corresponding to a point in the three-dimensional space to the camera plane. In stereoscopic video, in order to obtain the video information corresponding to the viewpoint that the user expects to watch through virtual viewpoint rendering, we need to obtain the depth map corresponding to each video frame. Therefore, a stereoscopic video sequence usually includes multiple channels of color video information and a depth map sequence corresponding to each channel of color video. Since the depth information of each pixel in the depth map is an integer value between 0 and 255, the depth map can be regarded as a video sequence composed of a series of grayscale images. In order to effectively store and transmit massive stereoscopic video data, stereoscopic video systems usually use multi-view video coding methods to compress multi-view video and depth map sequences respectively. Through intra-view and inter-view predictive coding, multi-view video coding methods can effectively compress the redundancy between multiple depth maps.

The traditional multi-view video coding method is a multi-view video coding extension scheme based on the H.264/MPEG-4 coding standard. In traditional multi-view video coding, the encoder codes each frame of image in units of macroblocks. Each 16×16 macroblock can be further divided into 16×8, 8×16 and 8×8 sub-blocks and 8×4, 4×8 and 4×4 sub-blocks. The different sub-blocks and sub-blocks are called prediction units. During the encoding process, the encoder performs motion estimation on each prediction unit to obtain the rate-distortion cost of motion compensation prediction. Then, mode selection based on rate-distortion optimization is performed to obtain the optimal coding mode and prediction residual for each block, and transform coding is performed on the residual. Experiments show that traditional multi-view video coding can achieve better compression performance on depth maps.

Different from traditional color video, the depth map only contains the distance information of each pixel and does not contain any video texture information. Therefore, the coded macroblocks inside the object in the depth map only contain a uniform depth texture, and the coded macroblocks on the edge of the object contain two or more discontinuous depth regions. Traditional macroblock division cannot effectively represent object edges. Especially under the condition of low coding rate, sub-blocks and sub-block patterns obtained by division are rarely selected. However, when encoding depth maps using traditional multi-view video coding methods, the encoder still needs to perform motion estimation and mode decisions for all macroblock modes. This process consumes a large amount of computing resources, which increases the complexity of the encoding side.

In order to obtain a higher depth image compression efficiency, it is necessary to design a more efficient encoding method according to the characteristics of the depth image. For this reason, the present invention proposes a depth map encoding method based on geometrical division, aiming at the feature that the depth map only contains object outline information. Better prediction results are obtained by adaptive geometric partitioning of depth macroblocks containing discontinuous motion fields. Thereby, the efficiency of depth map coding is improved, and the complexity of depth map coding is reduced at the same time.

At present, there are four patents related to the present invention that can be found, respectively disclosing a depth image encoding and compression method applied to 3DTV and FTV systems, a depth image encoding method in a stereoscopic television system, and a stereoscopic television system. A coding method for mid-depth images and a coding method for multi-view depth video, the application numbers are 200810063741.9, 200810161597.2, 200810120082.8 and 200910154138.6, respectively. Although the four mentioned patents are all related to the coding method of the depth map in stereoscopic video, the first, second and third patents only propose a method of using different quantization parameters for depth macroblocks in different regions. encoding scheme. The difference is that these three schemes design different depth macroblock classification methods; the fourth patent limits the coding modes of depth macroblocks in different areas, and sets more coding modes for depth macroblocks in object edge areas. mode, and fewer coding modes are set for depth macroblocks in non-object edge regions. However, the prediction units corresponding to all coding modes in this invention are still obtained based on the division method specified in the traditional H.264/MPEG-4 AVC coding standard. Therefore, none of the four mentioned patents deals with geometric partition-based coding methods for depth maps.

Contents of the invention

The present invention aims to solve at least one of the above-mentioned technical problems.

Therefore, an object of the present invention is to propose a depth map encoding method that can improve the compression efficiency of the depth map and reduce the coding complexity of the depth map.

Another object of the present invention is to provide a depth map encoding device.

In order to achieve the above object, the embodiment of the first aspect of the present invention proposes a depth map encoding method, including the following steps: establishing a plurality of division lines and forming a division set, wherein the plurality of division lines are used to wedge the depth macroblock Divide; encode the depth macroblock in intra-frame coding mode to obtain the corresponding first rate-distortion cost value; judge whether to code the depth macroblock in inter-frame coding mode, and if so, use inter-frame coding mode to encode all The depth macroblock is encoded to obtain the corresponding second rate-distortion cost value; continue to judge whether the depth macroblock contains a discontinuous motion vector field, if the depth macroblock is inter-coded and contains discontinuous motion vectors field, the depth macroblock is coded in a geometric partition coding mode, wherein the geometric partition coding mode further includes: selecting the optimal partition set from the partition set according to the brightness information of all pixels of the depth macroblock The division line divides the depth macroblock to obtain first and second depth sub-regions, and respectively performs motion estimation and predictive coding on the first and second depth sub-regions to obtain corresponding third rate-distortion cost values ; and comparing the rate-distortion cost values of the depth macroblocks in different coding modes, so as to select the coding mode with the smallest rate-distortion cost according to the comparison result to encode the depth macroblocks.

In addition, the depth map coding method according to the above-mentioned embodiments of the present invention may also have the following additional technical features:

In some examples, the set of partitions is:

G={ξ(ρ _i ,θ _i )|i=1,2,…,L},

Wherein, ξ(ρ _i , θ _i ) is the i-th dividing line in the set of dividing lines, ρ _i is the distance from the depth macroblock to the dividing line, and θ _i is the angle between the dividing line and the first direction , L is the number of dividing lines.

In some examples, the distances from the depth macroblock to the multiple division lines are all between [0, 8), and the angles between the multiple division lines and the first direction are all between [0, 2π).

In some examples, the step of judging whether the depth macroblock contains a discontinuous motion vector field further includes: acquiring brightness values of all pixels in the depth macroblock; calculating brightness values of all pixels in the depth macroblock and comparing the variance with a preset threshold, and if the variance is larger than the preset threshold, it is judged that the depth macroblock contains a discontinuous motion vector field.

In some examples, a step of selecting an optimal division line from the division set according to the brightness information of all pixels of the depth macroblock, and dividing the depth macroblock to obtain the first and second depth sub-regions It further includes: dividing the depth macroblock with each division line in the division set, and calculating the average luminance values of all the pixels in the two divided parts to obtain the corresponding first part and the second part respectively. The average luminance value of the two parts; the difference is obtained by making a difference between the average luminance values of the first part and the second part corresponding to each dividing line; and obtaining the dividing line that maximizes the absolute value of the difference as the optimal dividing line , to divide the depth macroblock to obtain first and second depth sub-regions.

In some examples, the step of encoding the depth macroblock in the intra-frame encoding mode to obtain the corresponding first rate-distortion cost value includes: according to the intra-frame encoding mode in the H.264/MPEG-4 AVC encoding standard The depth macroblock is intra-frame coded, and the corresponding optimal rate-distortion cost is used as the first rate-distortion cost value; the depth macroblock is coded in an inter-frame coding mode to obtain a corresponding second rate-distortion cost The cost value step includes: performing inter-mode coding on the depth macroblock according to the inter-frame coding mode in the H.264/MPEG-4AVC coding standard, and using the corresponding optimal rate-distortion cost as the second rate-distortion cost cost value.

In some examples, the step of comparing the rate-distortion cost values of the depth macroblocks in different coding modes, and selecting the coding mode with the smallest rate-distortion cost according to the comparison result to encode the depth macroblocks includes: if it is judged that the If the depth macroblock is inter-coded, then only the depth macroblock is intra-coded; if it is judged that the depth macroblock is inter-coded and does not contain discontinuous motion vector fields, then according to the first and second rates The comparison result of the distortion cost value selects the inter-frame coding mode or the intra-frame coding mode corresponding to the smaller rate-distortion cost value to encode the depth macroblock; and if it is judged that the depth macroblock is inter-frame coding and contains discontinuity motion vector field, then select the inter-frame coding mode, intra-frame coding mode or geometric partition coding mode corresponding to the smallest rate-distortion cost value to encode the depth macroblock according to the comparison results of the first to third rate-distortion cost values .

The embodiment of the second aspect of the present invention proposes a depth map encoding device, including: a division line establishment module, configured to establish a plurality of division lines and form a division set, wherein the plurality of division lines are used to wedge a depth macroblock Division; an intra-frame coding module, used to code the depth macroblock in an intra-frame coding mode to obtain a corresponding first rate-distortion cost value; an inter-frame coding judging module, used to judge whether to use an inter-frame coding mode to code the depth The macroblock is coded; the inter-frame coding module is used to code the depth macroblock in the inter-frame coding mode to obtain the corresponding The second rate-distortion cost value; the geometric division judgment module is used to judge whether the depth macroblock contains a discontinuous motion vector field; the geometric division coding module is used to judge whether the depth macroblock contains in the geometric division judgment module In the case of a discontinuous motion vector field, the depth macroblock is coded in a geometric partition coding mode, wherein the geometric partition coding mode includes: according to the brightness information of all pixels of the depth macroblock, from the partition set Select the optimal dividing line to divide the depth macroblock to obtain the first and second depth sub-regions, and perform motion estimation and predictive coding on the first and second depth sub-regions respectively to obtain the corresponding third A rate-distortion cost value; and a coding mode selection module, used to compare the rate-distortion cost values of the depth macroblock in different coding modes, so as to select the coding mode with the smallest rate-distortion cost according to the comparison result so as to carry out the depth macroblock coding.

In addition, the depth map encoding device according to the above-mentioned embodiments of the present invention may also have the following additional technical features:

In some examples, the set of partitions is:

G={ξ(ρ _i ,θ _i )|i=1,2,…,L},

Wherein, ξ(ρ _i , θ _i ) is the i-th dividing line in the set of dividing lines, ρ _i is the distance from the depth macroblock to the dividing line, and θ _i is the angle between the dividing line and the first direction , L is the number of dividing lines.

In some examples, the distances from the depth macroblock to the multiple division lines are all between [0, 8), and the angles between the multiple division lines and the first direction are all between [0, 2π).

In some examples, the geometric division judging module is used to obtain the luminance values of all pixels in the depth macroblock, calculate the variance of the luminance values of all pixels in the depth macroblock, and compare the variance and If the variance is greater than the preset threshold, it is determined that the depth macroblock contains a discontinuous motion vector field.

In some examples, the geometric partition encoding module is configured to respectively divide the depth macroblock with each dividing line in the division set, and respectively calculate the average value of all pixels in the two divided parts respectively. Brightness value to obtain the corresponding average brightness value of the first part and the second part, respectively make a difference between the average brightness values of the first part and the second part corresponding to each dividing line to obtain the difference, and obtain the maximum absolute value of the difference The dividing line of is used as the optimal dividing line, so as to divide the depth macroblock to obtain the first and second depth sub-regions.

In some examples, the intra-frame coding module is configured to perform intra-frame coding on the depth macroblock according to the intra-frame coding mode in the H.264/MPEG-4 AVC coding standard, and use the corresponding optimal rate-distortion cost As the first rate-distortion cost value; the inter-frame coding module is used to perform inter-mode coding on the depth macroblock according to the inter-frame coding mode in the H.264/MPEG-4 AVC coding standard, and correspond to The optimal rate-distortion cost of is used as the second rate-distortion cost value. In some examples, the encoding mode selection module is configured to only select an intra encoding mode when the depth macroblock is not inter-encoded, and the depth macroblock is inter-encoded and does not contain discontinuous motion In the vector field, according to the comparison result of the first and second rate-distortion cost values, select the inter-frame coding mode or the intra-frame coding mode corresponding to the smaller rate-distortion cost value, and perform inter-frame coding on the depth macroblock and include not For continuous motion vector fields, the inter-frame coding mode, intra-frame coding mode or geometric partition coding mode corresponding to the smallest rate-distortion cost value is selected according to the comparison results of the first to third rate-distortion cost values.

According to the depth map coding method and device of the embodiments of the present invention, by designing a depth map coding method based on geometric partitioning, the best coding mode pair is selected from the inter-frame coding mode, intra-frame coding mode or geometric partition coding mode in a reasonable manner. The depth macroblocks are coded, thereby not only improving the compression efficiency of the depth map, reducing the compression time, but also reducing the complexity of coding the depth map, and realizing efficient coding of the depth map.

Additional aspects and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.

Description of drawings

The above and/or additional aspects and advantages of the present invention will become apparent and comprehensible from the description of the embodiments in conjunction with the following drawings, wherein:

FIG. 1 is a flowchart of a depth map encoding method according to an embodiment of the present invention;

Fig. 2 is a principle diagram of dividing a depth macroblock by dividing lines of a depth map encoding method according to an embodiment of the present invention;

FIG. 3 is a detailed flowchart of a depth map encoding method according to an embodiment of the present invention;

Fig. 4 is a working principle diagram of an encoder with a geometric partition encoding mode used in a depth map encoding method according to an embodiment of the present invention;

FIG. 5 is a prediction structure diagram of a depth macroblock prediction coding mode in which the coding method for a depth map according to an embodiment of the present invention uses the coder shown in FIG. 4 to code; and

Fig. 6 is a structural diagram of a depth map encoding device according to an embodiment of the present invention.

Detailed ways

Embodiments of the present invention are described in detail below, examples of which are shown in the drawings, wherein the same or similar reference numerals designate the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the figures are exemplary only for explaining the present invention and should not be construed as limiting the present invention.

In describing the present invention, it should be understood that the terms "center", "longitudinal", "transverse", "upper", "lower", "front", "rear", "left", "right", " The orientations or positional relationships indicated by "vertical", "horizontal", "top", "bottom", "inner" and "outer" are based on the orientations or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and Simplified descriptions, rather than indicating or implying that the device or element referred to must have a particular orientation, be constructed and operate in a particular orientation, and thus should not be construed as limiting the invention. In addition, the terms "first" and "second" are used for descriptive purposes only, and should not be understood as indicating or implying relative importance.

In the description of the present invention, it should be noted that unless otherwise specified and limited, the terms "installation", "connection" and "connection" should be understood in a broad sense, for example, it can be a fixed connection or a detachable connection. Connected, or integrally connected; it may be mechanically connected or electrically connected; it may be directly connected or indirectly connected through an intermediary, and it may be the internal communication of two components. Those of ordinary skill in the art can understand the specific meanings of the above terms in the present invention in specific situations.

In order to verify the encoding method of the embodiment of the present invention, in the following examples of the present invention, the test sequence of stereoscopic video encoding adopts the standard test sequence of "Book Arrival" with the name of SD format, and the resolution of this SD format video sequence is 1024×768, the test sequence contains 16 viewpoint videos and corresponding depth maps. In the following examples of the present invention, the depth map in the test sequence is used for testing, and the depth map sequence of the eighth viewpoint of the sequence is used as the test sequence. The sequence contains 100 frames of depth maps. As shown in Figure 4, the codec adopts the H.264/MPEG-4 MVC standard reference software JMVC 6.0; the number of frames of the encoder GOP (Group of Pictures, Group of Pictures) is 8; the temporal domain predictive encoding of the encoding adopts Hierarchical B The prediction structure of Picture (hierarchical bidirectional predictive coding frame, referred to as hierarchical B frame), the coding prediction structure diagram is shown in Figure 5. Each coded depth macroblock has a size of 16x16. The precision of the motion estimation is quarter-pixel precision. Other encoding parameters are set with the parameters specified in the main profile (MainProfile) of the H.264/MPEG-4 AVC standard.

A method for encoding a depth map according to an embodiment of the present invention will first be described below with reference to the accompanying drawings.

Referring to Fig. 1, the depth map encoding method according to the embodiment of the present invention includes the following steps:

Step S101 , establishing a plurality of division lines and forming a division set, wherein the plurality of division lines are used for wedge-shaped division of depth macroblocks. In one embodiment of the present invention, the division set is: G={ξ(ρ _i ,θ _i )|i=1,2,...,L}, where ξ(ρ _i ,θ _i ) is the division line set The i-th dividing line in , ρ _i is the distance from the depth macroblock to the dividing line, θ _i is the angle between the dividing line and the first direction, and L is the number of dividing lines.

因此，划分集合G={ξ(ρ_i,θ_i)|i＝1,2,…,L}中共包含256条可选的划分线，即L=256。As shown in FIG. 2 , a division set composed of multiple division lines defining a wedge-shaped geometric division is defined. The dividing line is defined as a straight line (the dividing line shown in FIG. 2 ), which is represented by the distance ρ to the center point of the depth macroblock and the angle θ between the straight line and the y-axis, that is, the first direction. In this embodiment, the distances between the multiple dividing lines are all located between [0, 8), and the angles between the multiple dividing lines and the first direction are all located between [0, 2π), that is, the two points of each dividing line The value ranges of parameters ρ and θ are [0,8) and [0,2π), the sampling interval of parameter ρ is 1, and its value set is ρ={0,1,2,…,7}, the parameter The sampling interval of θ is π/16, and its value set is

Therefore, the division set G={ξ(ρ _i ,θ _i )|i=1, 2, . . . , L} contains 256 optional division lines, ie L=256.

Step S102, encode the depth macroblock in an intra-frame coding mode to obtain a corresponding first rate-distortion cost value.

Specifically, the depth macroblock is intra-coded according to the intra-frame coding mode in the H.264/MPEG-4 AVC coding standard, and the corresponding optimal rate-distortion cost is used as the first rate-distortion cost value. More specifically, the same method as the intra-frame coding mode defined in the traditional H.264/MPEG-4 AVC coding standard is used for coding to perform intra-frame prediction coding on the depth macroblock, denoted as B _k . In this embodiment, the optimal rate-distortion cost of the current depth macroblock in the intra prediction coding mode is $J_k^{\mathrm{intra}}=5741,$ That is, the first rate-distortion cost is $J_k^{\mathrm{intra}}=5741.$

Step S103, judging whether the depth macroblock is coded in the inter-frame coding mode, and if yes, the depth macroblock is coded in the inter-frame coding mode to obtain a corresponding second rate-distortion cost value.

Specifically, perform inter-mode coding on the depth macroblock according to the inter-frame coding mode in the H.264/MPEG-4 AVC coding standard, and use the corresponding optimal rate-distortion cost as the second rate-distortion cost value . In this example, the above-mentioned inter-frame coding mode is a conventional inter-frame coding mode.

Step S104, continue to judge whether the depth macroblock contains a discontinuous motion vector field, if the depth macroblock is inter-coded and contains a discontinuous motion vector field, then encode the depth macroblock in the geometric partition coding mode, wherein, the geometric The division coding mode further includes: selecting an optimal division line from the division set according to the luminance information of all pixels of the depth macroblock to divide the depth macroblock to obtain the first and second depth sub-regions, and the first and second depth sub-regions respectively Motion estimation and predictive coding are performed on the two-depth sub-regions to obtain corresponding third rate-distortion cost values.

Specifically, in some examples, the step of judging whether to encode the depth macroblock in a geometric division mode, that is, judging whether the depth macroblock contains a discontinuous motion vector field includes:

其中N代表该深度宏块中像素点的个数，在本实施例中，N=256。假设该深度宏块B_k为当前帧的第529个深度宏块。设B_k的全部像素点的亮度值为 $\left\{Y_n^k\right|n=\mathrm{1,2,3},\·\·\·,N\},$ 其中，

的数值如下述矩阵所示：1. Obtain the brightness values of all pixels in the depth macroblock. For example, the set of all pixels of the currently coded depth macroblock B _k is

Where N represents the number of pixels in the depth macroblock, and in this embodiment, N=256. Assume that the depth macroblock B _k is the 529th depth macroblock of the current frame. Let the brightness value of all pixels of B _k be $\left\{Y_{\mathrm{no}}^k\right|\mathrm{no}=\mathrm{1,2,3},\·\·\·,N\},$ in,

The values of are shown in the following matrix:

$\left[\begin{array}{cccccccccccccccc}\mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}\\ \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}\\ \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}\\ \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}\\ \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}\\ \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 108</span>}\\ \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 108</span>}& \mathrm{<span\; class="notranslate">\; 108</span>}\\ \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 108</span>}& \mathrm{<span\; class="notranslate">\; 108</span>}& \mathrm{<span\; class="notranslate">\; 108</span>}\\ \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 108</span>}& \mathrm{<span\; class="notranslate">\; 108</span>}& \mathrm{<span\; class="notranslate">\; 108</span>}\\ \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 108</span>}& \mathrm{<span\; class="notranslate">\; 108</span>}\\ \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}\\ \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}\\ \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}\\ \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 108</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}\\ \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 108</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 108</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}\\ \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 38</span>}& \mathrm{<span\; class="notranslate">\; 108</span>}& \mathrm{<span\; class="notranslate">\; 108</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}& \mathrm{<span\; class="notranslate">\; 105</span>}\end{array}\right]$

2. Calculate the variance of the luminance values of all pixels in the depth macroblock. It can be known through calculation that the variance of the luminance values of all pixels in the depth macroblock Bk is 838.

3. Compare the magnitude of the variance with the preset threshold, and if the variance is greater than the preset threshold, it is determined that the depth macroblock contains a discontinuous motion vector field. For example, if the preset threshold is T _S =100, then the variance of the luminance values of all pixels in B _k is greater than T _S , so it is determined that the depth macroblock contains a discontinuous motion vector field.

Selecting an optimal dividing line from the dividing set according to the luminance information of all pixels of the depth macroblock to divide the depth macroblock to obtain the first and second depth sub-regions further includes:

1. Divide the depth macroblock with each division line in the division set, and calculate the average luminance values of all the pixels in the two divided parts respectively to obtain the corresponding average luminance values of the first part and the second part . For example, the encoder searches for the optimal dividing line corresponding to B _k by traversing each dividing line in the set G. During the traversal process, for each dividing line ξ(ρ _i ,θ _i ), divide B _k into two parts. Calculate the average luminance values of all pixels in the two parts, for example, when the values of the parameters ρ and θ are both 0, the division line is a vertical line segment at the center of the macroblock B _k . The division line divides the macroblock B _k into two parts, P ¹ (ρ ₀ ,θ ₀ ) and P ² (ρ ₀ ,θ ₀ ), which are equal in size on the left and right. At this time, the average luminance value of all pixels in the area P ¹ (ρ ₀ , θ ₀ ) is 40.2; the average luminance value of all pixels in the area P ² (ρ ₀ , θ ₀ ) is 68.1.

2. The difference is obtained by making a difference between the average luminance values of the first part and the second part corresponding to each dividing line. According to the example in step 1, the difference between the average brightness values between P ¹ (ρ ₀ ,θ ₀ ) and P ² (ρ ₀ ,θ ₀ ) is: d=40.2-68.1=-27.9

3. Obtain a dividing line that maximizes the absolute value of the difference as the optimal dividing line, so as to divide the depth macroblock to obtain the first and second depth sub-regions. According to the example in step 2, the absolute value of the difference is: Δd=|40.2-68.1|=27.9.

In the same way, for all the division lines in the division line set G, the corresponding depth division difference after division is obtained in the manner of steps 1-3, and the division line with the largest depth division difference is taken as the optimal value of the current macroblock. The dividing line, the optimal dividing line is denoted as ξ(ρ ^opt ,θ ^opt ). In this embodiment, the maximum depth division difference for the depth macroblock B _k is Δd _max =65.6. The parameters of the corresponding optimal dividing line ξ(ρ ^opt ,θ ^opt ) are respectively ρ ^opt =4,

According to the above example, perform motion estimation on the first and second depth sub-regions to obtain the corresponding third rate-distortion cost value, specifically: use the optimal dividing line ξ(ρ ^opt ,θ ^opt ) to divide B _k into first and second depth sub-regions, denoted as P ¹ (ρ ^opt ,θ ^opt ) and P ² (ρ ^opt ,θ ^opt ). Then perform motion estimation on the first and second depth sub-regions respectively, and calculate their corresponding rate-distortion costs. In this example, the third rate-distortion cost of B _k in the geometric partition coding mode is

${\mathrm{<span\; class="notranslate">\; J</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}^{\mathrm{<span\; class="notranslate">\; GEOs</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1438</span>}<span\; class="notranslate">\; .</span>$

Step S105 , comparing the rate-distortion cost values of the depth macroblocks in different coding modes, so as to encode the depth macroblocks in the coding mode with the smallest rate-distortion cost according to the comparison result. Specifically, it includes the following steps:

1. If it is determined that inter-coding is not performed on the depth macroblock, only intra-coding is performed on the depth macroblock.

2. If it is judged that the depth macroblock is inter-frame coded and does not contain a discontinuous motion vector field, then select the inter-frame coding mode corresponding to the smaller rate-distortion cost value according to the comparison result of the first and second rate-distortion cost values or Intra coding mode codes depth macroblocks.

3. If it is judged that the depth macroblock is inter-coded and contains discontinuous motion vector fields, then select the inter-frame coding mode and intra-frame coding mode corresponding to the smallest rate-distortion cost value according to the comparison results of the first to third rate-distortion cost values. A coding mode or a geometric partition coding mode encodes depth macroblocks.

As a specific example, take the above-mentioned B _k as an example: the rate-distortion optimization-based mode selection is performed on B _k . In this embodiment, since the encoder predicts the geometric partition mode of _Bk , it is necessary to select the mode with the smallest rate-distortion cost among the three modes of geometric partition coding mode, inter-frame coding mode, and intra-frame coding mode as The optimal coding mode for B _k . For B _k , the minimum rate-distortion penalty is: $\min (J_k^{\mathrm{GEOs}},J_k^{\mathrm{inter}},J_k^{\mathrm{intra}})=J_k^{\mathrm{GEOs}}=1438.$

Therefore, the optimal coding mode of B _k is the geometric partition coding mode. Then the B _k is coded in the geometric partition coding mode.

Referring to FIG. 3, as a specific example, the depth map encoding method in the embodiment of the present invention can be performed according to the following steps:

Step S110, setting a set of dividing lines in the geometrical dividing mode.

Step S120, start encoding the current macroblock B _k .

Step S130, judging whether B _k is an intra-coded macroblock, if yes, go to step S140, otherwise go to step S220.

Step S140, perform intra-frame predictive coding on B _k and calculate the rate-distortion cost in this mode, and turn to step S150.

Step S150, judging whether B _k is encoded in the inter-frame mode, if yes, go to step S160, otherwise go to step S200.

In step S160, encode B _k in a conventional inter-frame prediction mode and calculate the rate-distortion cost in this mode, and proceed to step S170.

Step S170, judging whether B _k contains a discontinuous motion field. If yes, go to step S180, otherwise go to step S200.

Step S180, searching for the corresponding optimal dividing line according to the brightness information of each pixel in _Bk .

Step S190, perform motion estimation and encoding on the two parts divided by the optimal dividing line, and calculate the corresponding rate-distortion cost.

Step S200, select the mode with the smallest rate-distortion cost as the optimal coding mode of B _k .

Step S210, using the optimal mode to encode B _k .

Step S220, judging whether B _k is the last macroblock of the current frame. If yes, the encoding ends; otherwise, go to step S120 to encode the next depth macroblock.

Referring to FIG. 6 , a further embodiment of the present invention proposes a depth map encoding device 600, including a division line establishment module 610, an intra-frame encoding module 620, an inter-frame encoding judgment module 630, an inter-frame encoding module 640, and a geometric division judgment module 650. , a geometric partition encoding module 660 and an encoding mode selection module 670. in:

The dividing line establishment module 610 is used to establish a plurality of dividing lines and form a division set, wherein the plurality of dividing lines are used for wedge-shaped division of depth macroblocks. The intra-frame coding module 620 is used for coding the depth macroblock in an intra-frame coding mode to obtain a corresponding first rate-distortion cost value. The inter-frame coding judging module 630 is used for judging whether to code the depth macroblock in an inter-frame coding mode. The inter coding module 640 is configured to code the depth macroblock in the inter coding mode to obtain the corresponding second rate-distortion cost when the inter coding judging module 630 determines to code the depth macro block in the inter coding mode. The geometric division judging module 650 is used for judging whether the depth macroblock contains discontinuous motion vector fields. The geometric partition coding module 660 is used to code the depth macroblock in the geometric partition coding mode when the geometric partition judging module 650 judges that the depth macroblock contains a discontinuous motion vector field, wherein the geometric partition coding mode includes: according to the depth macroblock The luminance information of all pixels in the division set is selected from the division set to select the optimal division line to divide the depth macroblock to obtain the first and second depth sub-regions, and perform motion estimation and Predictive encoding to obtain a corresponding third rate-distortion cost value. The encoding mode selection module 670 is used to compare the rate-distortion cost values of the depth macroblocks in different encoding modes, so as to select the encoding mode with the smallest rate-distortion cost according to the comparison result to encode the depth macroblocks.

In some examples, the set is partitioned as:

G={ξ(ρ _i ,θ _i )|i=1,2,…,L},

Wherein, ξ(ρ _i , θ _i ) is the i-th dividing line in the set of dividing lines, ρ _i is the distance from the depth macroblock to the dividing line, and θ _i is the angle between the dividing line and the first direction , L is the number of dividing lines.

In some examples, the distances from the depth macroblock to the multiple dividing lines are all between [0, 8), and the angles between the multiple dividing lines and the first direction are all between [0, 2π).

In some examples, the geometric division judging module 650 is used to obtain the brightness values of all pixels in the depth macroblock, and calculate the variance of the brightness values of all pixels in the depth macroblock, and compare the size of the variance with the preset threshold, if If the variance is greater than the preset threshold, it is determined that the depth macroblock contains a discontinuous motion vector field.

Further, the geometric partition encoding module 660 is used to divide the depth macroblock with each division line in the division set, and calculate the average luminance values of all the pixels in the two divided parts to obtain Get the corresponding average luminance values of the first part and the second part, respectively make a difference between the average luminance values of the first part and the second part corresponding to each dividing line, and obtain the dividing line that maximizes the absolute value of the difference As the optimal division line, the depth macroblock is divided to obtain the first and second depth sub-regions.

In some examples, the intra-frame coding module 620 is configured to perform intra-frame coding on the depth macroblock according to the intra-frame coding mode in the H.264/MPEG-4 AVC coding standard, and use the corresponding optimal rate-distortion cost as The first rate-distortion cost value.

In some examples, the inter-frame coding module 640 is configured to perform inter-frame coding on the depth macroblock according to the inter-frame coding mode in the H.264/MPEG-4 AVC coding standard, and use the corresponding optimal rate-distortion cost as the second rate-distortion cost value.

In some examples, the encoding mode selection module 670 is configured to only select an intra encoding mode when the depth macroblock is not inter-encoded, and the depth macroblock is inter-encoded and does not contain discontinuous motion vectors In the frame time, according to the comparison result of the first and second rate-distortion cost values, select the inter-frame coding mode or the intra-frame coding mode corresponding to the smaller rate-distortion cost value, and perform inter-frame coding on the depth macroblock and include discontinuity When the motion vector field is selected, according to the comparison results of the first to third rate-distortion cost values, select the inter-frame coding mode, intra-frame coding mode or geometric partition coding mode corresponding to the smallest rate-distortion cost value so as to carry out the depth macroblock coding.

According to the depth map coding method and device of the embodiments of the present invention, by designing a depth map coding method based on geometric partitioning, the best coding mode pair is selected from the inter-frame coding mode, intra-frame coding mode or geometric partition coding mode in a reasonable manner. The depth macroblocks are coded, thereby not only improving the compression efficiency of the depth map, reducing the compression time, but also reducing the complexity of coding the depth map, and realizing efficient coding of the depth map.

In the description of this specification, descriptions referring to the terms "one embodiment", "some embodiments", "example", "specific examples", or "some examples" mean that specific features described in connection with the embodiment or example , structure, material or characteristic is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

Although the embodiments of the present invention have been shown and described, those skilled in the art can understand that various changes, modifications, substitutions and modifications can be made to these embodiments without departing from the principle and spirit of the present invention. The scope of the invention is defined by the claims and their equivalents.

Claims (14)

1. A depth map coding method, comprising:

establishing a plurality of dividing lines and forming a dividing set, wherein the dividing lines are used for carrying out wedge-shaped division on the depth macro block;

coding the depth macro block in an intra-frame coding mode to obtain a corresponding first rate distortion cost value;

judging whether the depth macro block is coded in an inter-frame coding mode, if so, coding the depth macro block in the inter-frame coding mode to obtain a corresponding second rate distortion cost value;

continuing to determine whether the depth macroblock contains a discontinuous motion vector field, and if the depth macroblock contains a discontinuous motion vector field and is inter-coded, coding the depth macroblock in a geometric partition coding mode, wherein the geometric partition coding mode further comprises: selecting an optimal dividing line from the dividing set according to the brightness information of all pixel points of the depth macro block to divide the depth macro block to obtain a first depth sub-area and a second depth sub-area, and respectively performing motion estimation and predictive coding on the first depth sub-area and the second depth sub-area to obtain a corresponding third rate distortion cost value; and

and comparing the rate distortion cost values of the depth macro blocks under different coding modes, and selecting the coding mode with the minimum rate distortion cost according to the comparison result to code the depth macro blocks.

2. The method of claim 1, wherein the partition set is:

G={ξ(ρ_i,θ_i)|i＝1,2,…,L}，

where ξ (ρ)_i,θ_i) Is the ith dividing line, rho, in the dividing line set_iIs the distance, θ, of the depth macroblock to the dividing line_iIs the angle between the dividing line and the first direction, and L is the number of the dividing lines.

3. The method of claim 2, wherein distances from the depth macroblock to the plurality of dividing lines are all between [0,8), and angles from the first direction are all between [0,2 π).

4. The method of claim 1, wherein the step of determining whether the depth macroblock contains a discontinuous motion vector field further comprises:

acquiring the brightness values of all pixel points in the depth macro block;

calculating the variance of the brightness values of all pixel points in the depth macro block; and

and comparing the variance with a preset threshold, and if the variance is greater than the preset threshold, judging that the depth macro block contains a discontinuous motion vector field.

5. The depth map encoding method of claim 1, wherein the step of selecting an optimal partition line from the partition set to partition the depth macroblock into the first and second depth sub-regions according to luminance information of all pixel points of the depth macroblock further comprises:

dividing the depth macro block by each dividing line in the dividing set, and calculating the average brightness value of all pixel points in the two divided parts respectively to obtain the average brightness value of the corresponding first part and the second part;

respectively carrying out difference on the average brightness values of the first part and the second part corresponding to each dividing line to obtain a difference value; and

obtaining a dividing line that maximizes an absolute value of the difference as the optimal dividing line to divide the depth macroblock to obtain first and second depth sub-regions.

6. The method of claim 1, wherein the step of encoding the depth macroblock in intra coding mode to obtain the corresponding first rate-distortion cost value comprises:

carrying out intra-frame coding on the depth macro block according to an intra-frame coding mode in an H.264/MPEG-4AVC coding standard, and taking the corresponding optimal rate distortion cost as the first rate distortion cost value;

the step of encoding the depth macroblock in the inter-coding mode to obtain the corresponding second rate-distortion cost value comprises:

and performing inter-mode coding on the depth macro block according to an inter-coding mode in the H.264/MPEG-4AVC coding standard, and taking the corresponding optimal rate distortion cost as the second rate distortion cost value.

7. The method of claim 1, wherein the step of comparing the rate-distortion cost values of the depth macroblocks in different coding modes to select a coding mode with the lowest rate-distortion cost according to the comparison result to code the depth macroblock comprises:

if the depth macro block is judged not to be subjected to inter-frame coding, only the depth macro block is subjected to intra-frame coding;

if the depth macro block is judged to be inter-coded and does not contain discontinuous motion vector fields, selecting an inter-coding mode or an intra-coding mode corresponding to a smaller rate distortion cost value according to a comparison result of the first rate distortion cost value and the second rate distortion cost value to code the depth macro block; and

and if the depth macro block is judged to be inter-coded and comprises discontinuous motion vector fields, selecting an inter-coding mode, an intra-coding mode or a geometric partition coding mode corresponding to the minimum rate distortion cost value according to the comparison result of the first rate distortion cost value to the third rate distortion cost value to code the depth macro block.

8. A depth map encoding apparatus, comprising:

the dividing line establishing module is used for establishing a plurality of dividing lines and forming a dividing set, wherein the dividing lines are used for carrying out wedge-shaped division on the depth macro block;

the intra-frame coding module is used for coding the depth macro block in an intra-frame coding mode to obtain a corresponding first rate distortion cost value;

the inter-frame coding judging module is used for judging whether the depth macro block is coded in an inter-frame coding mode;

the inter-frame coding module is used for coding the depth macro block in the inter-frame coding mode to acquire a corresponding second rate distortion cost value when the inter-frame coding judging module judges that the depth macro block is coded in the inter-frame coding mode;

a geometric partitioning judgment module, configured to judge whether the depth macroblock includes a discontinuous motion vector field;

a geometric partition coding module, configured to code the depth macroblock in a geometric partition coding mode when the geometric partition determining module determines that the depth macroblock is inter-coded and contains a discontinuous motion vector field, where the geometric partition coding mode includes: selecting an optimal dividing line from the dividing set according to the brightness information of all pixel points of the depth macro block to divide the depth macro block to obtain a first depth sub-area and a second depth sub-area, and respectively performing motion estimation and predictive coding on the first depth sub-area and the second depth sub-area to obtain a corresponding third rate distortion cost value; and

and the coding mode selection module is used for comparing the rate distortion cost values of the depth macro block under different coding modes so as to select the coding mode with the minimum rate distortion cost according to the comparison result so as to code the depth macro block.

9. The depth map coding apparatus of claim 8, wherein the partition set is:

G={ξ(ρ_i,θ_i)|i＝1,2,…,L}，

where ξ (ρ)_i,θ_i) Is the ith dividing line, rho, in the dividing line set_iIs the distance, θ, of the depth macroblock to the dividing line_iIs the angle between the dividing line and the first direction, and L is the number of the dividing lines.

10. The depth map coding apparatus of claim 9, wherein distances from the depth macroblock to the plurality of division lines are each between [0,8), and angles from the first direction are each between [0,2 pi ].

11. The apparatus according to claim 8, wherein the geometric partition determining module is configured to obtain luminance values of all pixels in the depth macroblock, calculate a variance of the luminance values of all pixels in the depth macroblock, compare the variance with a preset threshold, and determine that the depth macroblock includes a discontinuous motion vector field if the variance is greater than the preset threshold.

12. The depth map encoding device of claim 8, wherein the geometric partition encoding module is configured to divide the depth macro block by each partition line in the partition set, calculate average luminance values of all pixel points in the two divided portions to obtain average luminance values of the corresponding first portion and second portion, difference the average luminance values of the first portion and second portion corresponding to each partition line to obtain a difference value, and obtain the partition line with the largest absolute value of the difference value as the optimal partition line to divide the depth macro block to obtain the first and second depth sub-regions.

13. The depth map coding apparatus of claim 8,

the intra-frame coding module is used for carrying out intra-frame coding on the depth macro block according to an intra-frame coding mode in an H264/MPEG-4 AVC coding standard and taking the corresponding optimal rate distortion cost as the first rate distortion cost value;

and the inter-frame coding module is used for carrying out inter-frame mode coding on the depth macro block according to an inter-frame coding mode in the H264/MPEG-4 AVC coding standard and taking the corresponding optimal rate distortion cost as the second rate distortion cost value.

14. The depth map encoding apparatus of claim 8, wherein the encoding mode selection module is configured to select only an intra-frame encoding mode when the depth macroblock is not inter-encoded, select an inter-frame encoding mode or an intra-frame encoding mode corresponding to a smaller rate-distortion cost value according to a comparison result between a first rate-distortion cost value and a second rate-distortion cost value when the depth macroblock is inter-encoded and does not include a discontinuous motion vector field, and select an inter-frame encoding mode, an intra-frame encoding mode, or a geometric partition encoding mode corresponding to a smallest rate-distortion cost value according to a comparison result between the first rate-distortion cost value and the third rate-distortion cost value when the depth macroblock is inter-encoded and includes a discontinuous motion vector field.

Images