论文部分内容阅读
针对立体视频中存在着大量的感知冗余,首先通过主观实验研究梯度和纹理因素对立体掩蔽效应的影响,建立双目恰可察觉编码失真(BJNCD)模型;同时考虑到人类视觉特性(HVS)的视觉敏感度并非恒定不变,当视网膜离心率变大时,像素的视觉阈值也随之变大,因此结合HVS的视网膜中心凹感知特性建立了基于中心凹的双目恰可察觉编码失真(FBJNCD)模型。为验证所提出模型的有效性,将其应用于多视点高效视频编码(MV-HEVC)测试平台中对立体视频进行非对称编码。实验结果表明,所提出模型在保持立体视频感知质量的同时能够有效节省编码码率,提高了立体视频压缩效率。
Considering the large amount of perceived redundancy in stereoscopic video, the binocular just perceptible coding distortion (BJNCD) model is first established by subjective experiments to study the effect of gradient and texture on the stereoscopic masking effect. At the same time, human visual characteristics (HVS) Of the visual sensitivity is not constant, when the retina eccentricity becomes larger, the visual threshold of the pixel also becomes larger, so the combination of the HVS retinal foveal perception characteristics based on the fovea binocular just detectable coding distortion FBJNCD) model. In order to verify the effectiveness of the proposed model, this method is applied to multi-view high-efficiency video coding (MV-HEVC) test platform for asymmetric stereo coding. The experimental results show that the proposed model can effectively reduce the coding bit rate while maintaining the perceived quality of stereoscopic video and improve the compression efficiency of stereoscopic video.