论文部分内容阅读
针对井下监控视频普遍亮度较低以及监控对大部分背景细节不关注的情况,提出了一种基于改进时空域JND模型的视频主观质量改善算法。该模型改进了纹理掩蔽效应与边缘掩蔽效应的结合方式,避免了同一区域可能出现的两种掩蔽效应的不一致;且该算法使用与空域JND相同的单元进行计算并使用相同的加权参数以避免环境噪声对单个象素造成的影响。根据改进的JND模型对于H.264帧间编码进行自适应的残差预处理以降低编码码率。实验结果表明,改进算法可在保持同等主观视觉质量的前提下,明显降低编码所需的码率,从而提高固定传输带宽下井下监控视频的传输路数及帧率。
Aimed at the low general brightness of video surveillance in underground and the lack of attention to most background details, a video subjective quality improvement algorithm based on improved JND model is proposed. This model improves the combination of texture masking and edge masking effects and avoids inconsistency between the two masking effects that may occur in the same region. The algorithm uses the same elements as the JND of the airspace to calculate and use the same weighting parameters to avoid the environment The impact of noise on a single pixel. According to the improved JND model, adaptive residual pre-processing for H.264 inter-frame coding is implemented to reduce the coding bit rate. The experimental results show that the improved algorithm can reduce the bit rate required for coding significantly while maintaining the same subjective visual quality, thereby increasing the number of transmission channels and the frame rate of underground monitoring video under a fixed transmission bandwidth.