Bayer图像的无损压缩算法及其硬件实现  

Hardware implementation of Bayer image lossless compression algorithm

作　　者：黄文俊 元辉 李富勇 魏锡彦 HUANG Wenjun;YUAN Hui;LI Fuyong;WEI Xiyan(Chuangrun Quantum Technology(Shanghai)Co.,Ltd.,Shanghai 201203,China;School of Control Science and Engineering,Shandong University,Jinan 250061,China;Shandong Branch of China Telecom Group System Integrated Co.,Ltd.,Jinan 250101,China;Shandong Branch of China Telecom Group Co.,Ltd.,Jinan 250101,China)

机构地区：[1]创润量子科技(上海)有限公司,上海201203 [2]山东大学控制科学与工程学院,山东济南250061 [3]中国电信集团系统集成有限责任公司山东分公司,山东济南250101 [4]中国电信集团有限公司山东分公司,山东济南250101

出　　处：《电子设计工程》2022年第11期183-188,共6页Electronic Design Engineering

摘　　要：Bayer格式是目前应用最广的图像传感器数据输出格式,在数码相机成像端直接对Bayer图像进行无损压缩可以显著降低后续处理芯片的内存消耗。主流图像压缩标准复杂度较高,并且对Bayer图像压缩性能一般,故此提出了一种适合于FPGA实现的无损编码方案。该方案将图像分割成的若干个子块作为编码单元,使用多种预测模式降低图像信息在空间上的冗余性。同时提出一种改进的自适应Huffman编码算法,根据已编码图像块中的预测残差信息动态调整Huffman树,降低了计算复杂度。测试结果表明,该方案在满足实时性的条件下以更少的资源消耗达到了主流无损压缩方案的压缩性能。The Bayer format is the most widely used image sensor output format,the lossless compression of Bayer image at the imaging end of digital camera can significantly reduce the memory consumption of subsequent processing device.Mainstream image compression technique have high complexity but their performance are mediocre for Bayer image.This paper proposed a lossless compression scheme that is suitable for FPGA.In this scheme,Bayer image is diveded into several blocks as coding units,and then we reduce the spatial redundancy of image information by mutiple prediction models.An improved adaptive Huffman coding algorithm with low complexity is proposed,which can adjust the Huffman tree according to the predicted residual in the encoded blocks.The test results indicate that the compression performance of proposed sheme is close to mainstream lossless compression shemes with low resource consumption and delay.

关 键 词：BAYER格式 无损压缩 图像预测 HUFFMAN编码 FPGA 

分 类 号：TN911.73[电子电信—通信与信息系统] TP391.41[电子电信—信息与通信工程]