DSP体系结构下视觉监控优化方法研究

Research on optimization methods for intelligent vision surveillance algorithm on DSP

探讨了智能视觉监控算法在TIDSP平台上的移植方法。详细介绍了算法优化、CCS编译器优化、C代码优化、汇编指令优化及存储器配置优化等优化技术。通过具体的监控实例,在TMS320DM642平台上实现了多个运动目标的实时检测、跟踪算法,结果表明,对DSP进行优化之后其性能获得了指数级的增长。

First,a methodology for migrating intelligent visual monitoring algorithms onto the TI DSP platform is discussed.Then, the methods for optimization, algorithms of optimization in C language and in assembly directive, and the technologies of optimization in memory configuration are illuminated in detail.System of surveillance for multi-objects has been carried into effect in TMS320DM642 DSP.The performance has stepped up after optimizing as a result.