快速高效无损图像压缩系统的低功耗硬件实现  

Low power hardware implementation of the fast and efficient lossless image compression system

作　　者：薛金勇[1] 黑勇[1] 陈黎明[1] 

机构地区：[1]中国科学院微电子研究所,北京100029

出　　处：《哈尔滨工程大学学报》2014年第3期368-372,共5页Journal of Harbin Engineering University

基　　金：国家863计划资助项目(2008AA010704)

摘　　要：针对快速高效无损图像压缩系统(FELICS)中,Golomb-Rice编码k参数选取增加系统复杂度和限制系统吞吐量的问题,依据基于上下文的图像压缩理论,采用JPEG-LS中的序列参数估计方法,研究了限长Golomb-Rice编码。给出了面向超大规模集成电路(VLSI)实现的低功耗快速高效无损图像压缩系统。VLSI-oriented FELICS简化了k参数的选取,提高了系统的吞吐率;针对贝尔图像格式的算法扩展提高了FELICS算法对Bayer图像的压缩效果。在SMIC 0.13μm工艺条件下,基于该方法完成了面向低功耗胃窥镜的VLSI设计。当系统工作在25 MHz,图像数据时钟为24 MHz时,编码每帧图像的功耗仅为11.15μW,VGA图像的吞吐率可以达到60 f/s。The Golomb-Rice code of the fast and efficient lossless image compression system （ FELICS） requires es-timation of the coding parameter value k, which increases the system complexity and limits the throughput of the system.This research into the limited-length Golomb-Rice code adopting sequential parameter estimation of JPEG-LS was made based on the context model of the image compression theory .The very large scale integration （ VLSI）-oriented low power FELICS algorithm was proposed .The VLSI-oriented low power FELICS algorithm simplified the selection of the coding parameter and improved the throughput of the system .The VLSI-oriented FELICS with Bayer expansion improved the compression ratio of the Bayer images .The VLSI design based on the proposed algorithm was implemented under the condition of the SMIC （ Semiconductor Manufacturing International Corporation ） 0.13 μm process.When the system works at 25 MHz clock frequency with a 24 MHz pixel clock frequency in VGA resolution, the power dissipation per frame is only 11.15μW, and the throughput of the system is up to 60 frames per second .

关 键 词：快速高效无损图像压缩系统 Golomb-Rice编码 吞吐量 超大规模集成电路 贝尔图像 低功耗 

分 类 号：TN47[电子电信—微电子学与固体电子学]