摘要
文章介绍 JPEG编码的 DSP实现 ,且着重讨论 DCT快速算法和 DCT系数量化的快速实现 ,将 8点 DCT分解为蝶形运算和乘法累加运算两级结构 ,利用 DSP的乘法累加指令和双字加 /减法指令快速实现 DCT。用小数乘法运算代替 DCT系数量化的除法运算 ,快速实现了量化操作 ,JPEG算法用汇编语言编程实现 ,编译后的可执行代码仅 2
The realization of JPEG(Joint Photographic Expert Group) coding based on the digital signal processor(DSP) is introduced, and the fast discrete cosine transform(DCT) algorithm and the DCT coefficients' quantization are analyzed in detail. The 8-point DCT is divided into two steps-the butterfly operation and the multiplication-accumulation operation, and it is realized fast with DSP's multiplication-accumulation instructions and double-word addition/subtraction instructions. The DCT coefficients' quantization is realized with multiplication instead of division. This JPEG coding program is formulated in assembler language, and the executive code length of the compiled program is only 2 048 words.
出处
《合肥工业大学学报（自然科学版）》
CAS
CSCD
2002年第4期531-534,共4页
Journal of Hefei University of Technology：Natural Science
