论文部分内容阅读
ＤＣＴ是一种被广泛应用于图像压缩编码中的正交变换。ＤＣＴ的运算量极大，因而在一些高速或实时场合，能否快速实现ＤＣＴ就成为一个关键因素。ＴＭＳ３２０Ｃ８０（ＭＶＰ）作为新一代可编程的通用数字信号处理芯片，它支持ＳＩＭＤ和ＭＩＭＤ两种并行模式。文中描述了ＤＣＴ的快速算法和ＴＭＳ３２０Ｃ８０的基本结构及特性，并详述了并行实现的策略，对两种并行模式的加速比作了比较，然后选用了一种高效算法，在ＴＭＳ３２０Ｃ８０上实现。并对结果进行了理论上的分析，证明能满足实时性的要求
DCT is an orthogonal transform that is widely used in image compression coding. DCT is extremely computationally intensive, so the ability to achieve DCT quickly becomes a key factor in some high-speed or real-time applications. TMS320C80 (MVP) as a new generation of programmable general purpose digital signal processing chip, it supports SIMD and MIMD two parallel modes. In this paper, the fast algorithm of DCT and the basic structure and characteristics of TMS320C80 are described. The strategy of parallel implementation is described in detail. The speedup ratio of two parallel modes is compared. Then an efficient algorithm is selected and implemented on TMS320C80. And the results of a theoretical analysis to prove that can meet the real-time requirements