采用灰度共生矩阵进行深度预判的3D-HEVC深度图帧内快速编码算法

Fast Depth Intra Coding using Gray Level Co-occurrence Matrix in 3D-HEVC

  • 摘要: 针对3D视频的3D-HEVC编码标准以多视点纹理视频和深度视频格式进行编码,其深度图编码仍延续纹理视频编码的模式和编码尺寸遍历选择,使得3D-HEVC的编码复杂度居高不下。本文针对深度图帧内预测编码,采用灰度共生矩阵对深度图中的CTU进行计算,统计并分析其矩阵中非零值个数与CTU分割深度的关系,根据非零值个数分布规律,设定阈值,使得帧内编码时可以预判编码模块的分割深度,从而选择性跳过部分不同深度CU的帧内预测过程。经过HTM16.0测试平台的检验,本算法在全帧内编码模式下,测试序列合成视点比特率仅增加0.08%的同时,平均节省了16.8%的编码时间,与其他同类较新算法在HTM16.0平台上的性能比较也有一定的优势。

     

    Abstract: As an extension of High Efficiency Video Coding for 3D video, 3D-HEVC supports the multi-view video plus depth (MVD) format, which consists not only the texture but also the depth map sequence for each view. To reduce the computational complexity of depth map coding, a fast algorithm for depth intra coding based on the gray level co-occurrence matrix is proposed in this paper. By studying the correlation of non-zero number in gray level co-occurrence matrix with CTU numbers in different partitioning depth, the CU size of intra coding can be prejudged, which means the intra prediction process of CU in other depth can be selectively skipped. The experimental results testing in HTM16.0 show that the proposed algorithm achieves an average 16.8% encoding time saving with a small synthesized total bitrate loss of 0.08% under all intra configuration. The proposed algorithm also superior to other state-of-the-art methods under HTM16.0 platform.

     

/

返回文章
返回