一种使用帧间差值的图像传感器片上视频压缩算法  

An On-Chip Video Compression Algorithm for Image Sensor Using Differences Between Frames

作　　者：蒋永唐 徐江涛 陈全民 衡佳伟 JIANG Yongtang;XU Jiangtao;CHEN Quanmin;HENG Jiawei(School of Microelectronics,Tianjin University,Tianjin 300072,China;Tianjin Key Laboratory of Imaging and Sensing Microelectronic Technology,Tianjin 300072,China)

机构地区：[1]天津大学微电子学院,天津300072 [2]天津市成像与感知微电子技术重点实验室,天津300072

出　　处：《传感技术学报》2021年第3期319-329,共11页Chinese Journal of Sensors and Actuators

基　　金：国家重点研发计划项目(2019YFB2204202)。

摘　　要：为了减少图像传感器视频数据的输出,提出了一种通过编码相邻两帧之间差值的无损视频压缩算法。算法首先将基于差分脉冲编码调制原理的差分操作在模拟域实现,减小了电路的复杂度。然后两帧之间的差值被基于块的无损压缩方案编码。实验结果证明,压缩后的图像数据可以被无损失的还原。通过对7个具有代表性的8位深度1280×720@60 fps的样本视频进行测试,在块大小为4×4和模式切换阈值为63时实现了最佳的压缩效果。在几乎没有光的条件下压缩率高达78.5%。在复杂运动场景下该算法压缩率为43.5%。提出的压缩算法更适用于长时间处于静止场景的视频录制。In order to reduce the output of video data from image sensor,a lossless video compression algorithm by transmitting the encoded differences between two consecutive frames is presented.The design is based on differential pulse code modulation(DPCM).The differential operations are performed in the analog domain to decrease the area complexity of the circuit.The proposed compression algorithm encodes the picture differences with the block-based lossless compression(BLC)coding scheme.Experimental results demonstrate that the video is restored without loss by decompressing the encoded data.By testing seven representative sample videos of 1280×720@60 fps with 8-bit depth,the best block size of 4×4 and the best threshold 63 of mode switching are obtained.The compression gain reached 78.5% in almost dark scenes.A compression gain of 43.5% is achieved in scene with complex content changes.The proposed algorithm is more suitable for the compression of long-time static scenes.

关 键 词：视频压缩 CMOS图像传感器 基于块无损压缩 差分脉冲编码调制 压缩率 

分 类 号：TN919.81[电子电信—通信与信息系统] TN919.31[电子电信—信息与通信工程]