基于视点相关性的光场图像压缩算法

为了探索虚拟绘制视点之间的强相关性,提高光场图像的压缩效率,提出一种基于视点相关性的光场图像压缩算法.该算法基于高清视频编码屏幕内容编码扩展平台,利用线性加权算法以及帧内块拷贝混合预测算法来提升编码块的预测精度;并利用率失真优化过程来自适应地选择最优的编码块大小以及预测模式.结果表明,所提算法相比于高清视频编码标准可以获得2.55dB的平均BD-峰值信噪比编码增益,同时可以获得较好的虚拟视点绘制质量.该算法充分利用虚拟绘制视点之间的强相关性,提高了光场图像的编码效率.

Light field image compression method based on correlation of rendered views

In order to explore the strong correlation between virtual rendering viewpoints and improve the compression efficiency of optical field image, an optical field image compression algorithm based on viewpoint correlation was proposed. The algorithm was based on high efficiency video coding (HEVC) and screen content coding expansion platform. Linear weighting algorithm and intra-block copy hybrid prediction algorithm were used to improve the prediction accuracy of coded blocks. Distortion optimization process was used to adaptively select the optimal block size and prediction mode. The experimental results show that, the average BJONTEGAARD delta peak signal-to-noise ratio(BD-PSNR) coding gain of the proposed algorithm is 2.55dB compared with that of HEVC standard. At the same time, better quality of virtual view rendering can be gotten. The algorithm makes full use of the strong correlation between virtual rendering viewpoints and improves the coding efficiency of optical field images.