This paper aims to describe the implementation of shadow suppression algorithm in an intelligent tracking system based on multimedia Digital Signal Processor (DSP). In our application, the fast filling and noise elimination algorithm is added to improve the accuracy of shadow suppression. In order to achieve real-time performance, several optimization techniques have been applied to shadow suppression. The algorithm level optimization simplifies the complexity of the algorithm. The program level and instruction level optimization are inspired by the characteristic of the processor chip. These methods might be general-propose for the MCUs or CPUs with pipeline feature. Experimental results show that the optimization greatly reduces the CPU load and contributes more time to the future work.
展开▼