摘要
在遥感图像处理过程中,通常需要对大画幅数字图像进行实时压缩,以便能通过有限的通用信道传送到地面进行判读处理,使用普通的视频压缩技术很难满足实用要求,采用多DSP+FPGA构建实时压缩平台已经成为主要的研究方向;在开发DSP压缩平台的过程中,一般使用C语言编译器进行软件开发,所得到的代码实时性较差;分别采用项目级、算法级和指令级三级优化的方法,对小波变换、量化和编码等程序进行优化,提高了压缩算法的执行速度;实验结果表明,对于4096×3072的图像,使用经过优化的压缩编码器,图像压缩速度达到了每秒6帧,这种优化方法具有一定的通用性。
In the process of data processing for remote sensing image, a big digital image is usually needed to be compressed in real time, so that it can be transmitted to the ground through limited universal channel for interpretation and processing. Because remote sensing image has weak local correlation, great data quantity, and abundance detail, the traditional video compression techniques can＇ t satisfy the practical need, so adopting multi--DSP-kFPGA （Digital Signal Processors and Field Programmable Gate Array） to construct a real--time compres- sion platform becomes a major study aspect. In the process of developing the DSP compression platform, C language compiler is usually ap- plied to developing the software, but the codes generated by the compiler are unqualified in real--time basis. This paper optimizes the pro- gram of Wavelet Transform, quantization and coding with the method of three--level optimization, including project--level, algorithm--level and instruction--level, and then increases the operation rate of compression arithmetic. The experimental results show that, for a image by 4096 × 3072, the rate of compression reaches 6 frames/second by using the optimized compression coder. The optimization method has a performance generality.
出处
《计算机测量与控制》
CSCD
北大核心
2010年第3期671-673,676,共4页
Computer Measurement &Control
基金
总装备部试验技术研究项目资助课题(2007SY4112002)
关键词
图像压缩
DSP
程序优化
image compression
DSP
program optimization
