嵌入式DSP平台纸币识别算法内存使用的优化

With the rapid development of computer technology, automation and intelligent systems are more widely used in all kinds of industry. Paper currency recognition system based on embeded platform will be more welcomed by SMEs.This paper introduces the optimization of memory usage in paper currency recognition system on embedded DSP platform TMS320 DM642.As a theoretical basis for optimizing memory usage, this paper describes the DM642 architecture at first; then shows the memory access bottleneck on DSP platform and the corresponding optimization methods; final illustrate the effect of optimizing memory usage by the algorithm and functions' efficiency.After the optimization of memory usage, the efficiency of the embeded paper currency recognition algorithm has been greatly improved.