适于航空面阵CCD相机的H.264编码帧内预测算法

发布时间：2018-02-12 21:11

本文关键词： 面帧CCD相机视频压缩 H.压缩最小平方预测　出处：《光电子.激光》2014年08期 　论文类型：期刊论文

【摘要】：为了提高H.264中帧内预测压缩编码效率,提出一种适于航空面阵CCD相机应用的H.264的新的帧内预测模式。其基本思想为:一个块中的每个像素预测值使用周围像素和权重采用一个N阶线性预测器计算得到,而权重值是由重构像素中与待编码像素的邻近块计算得到,可参考的邻近块使用l1范数评估。而权重值并不需要传输到解码器中而耗用大量时间。在H.264参考软件JM17.6上的实验结果表明,本文提出的帧内预测算法具有很好的率失真(RD)性能,与标准H.264帧内预测算法比较,对于图像而言,PSNR增加了0.57~0.17dB,比特率减少了9.71~3.94%;对于CIF格式视频序列,PSNR增加了0.05~0.14dB,比特率减少了1.23~1.52%;对于720p50格式视频序列,PSNR增加了0.01~0.05dB,比特率减少了0.25~0.47%。因此,本文算法非常适于航空面阵CCD相机的应用。
[Abstract]:In order to improve the coding efficiency of intra prediction compression in H.264, This paper presents a new intra prediction mode for H.264, which is suitable for aerial array CCD camera applications. The basic idea is that each pixel prediction value in a block is calculated using the surrounding pixels and the weights are calculated using a N order linear predictor. The weight value is calculated by the adjacent blocks of the reconstructed pixel and the pixel to be encoded, The reference adjacent blocks are evaluated using L1 norm, and the weight values do not need to be transmitted to the decoder to consume a lot of time. The experimental results on the H.264 reference software JM17.6 show that the proposed intra prediction algorithm has good rate-distortion (RDS) performance. Compared with the standard H.264 intra prediction algorithm, for the image, the PSNR increases by 0.57 / 0.17dB and the bit rate decreases by 9.71k3.94; for the CIF video sequence, the bit-rate is increased by 0.050.14dB, and the bit rate is reduced by 1.230.52; for the 720p50 video sequence, the bit-rate is reduced by 0.250.47dBby an increase of 0.01db; for the 720p50 video sequence, the bit-rate is increased by 0.01nb, and the bit rate is reduced by 0.250.47. therefore, for the video sequence in the 720p50 format, the bitrate decreases by 0.250.47. This algorithm is very suitable for the application of aerial array CCD camera.
【作者单位】：中国科学院长春光学精密机械与物理研究所;中国科学院研究生院;
【基金】：吉林省科技发展计划(20130522102JH)资助项目
【分类号】：TN919.81

【相似文献】