单片机与嵌入式系统应用
單片機與嵌入式繫統應用
단편궤여감입식계통응용
Microcontrollers & Embedded Systems
2015年
10期
33-36
,共4页
Mean Shift%硬件加速%实时处理%SOPC
Mean Shift%硬件加速%實時處理%SOPC
Mean Shift%경건가속%실시처리%SOPC
Mean Shift%hardware acceleration%real-time process%SOPC
为了解决Mean Shift算法基于浮点运算计算复杂、难以实时实现及跟踪窗固定的问题，提出一种适合在FPGA上实现的改进的Mean Shift算法。实验结果表明，本文提供的算法能自适应地更新跟踪窗的大小，在4 ms内实现对目标的稳定跟踪，程序执行效率比用DSP实现提高5倍左右。
為瞭解決Mean Shift算法基于浮點運算計算複雜、難以實時實現及跟蹤窗固定的問題，提齣一種適閤在FPGA上實現的改進的Mean Shift算法。實驗結果錶明，本文提供的算法能自適應地更新跟蹤窗的大小，在4 ms內實現對目標的穩定跟蹤，程序執行效率比用DSP實現提高5倍左右。
위료해결Mean Shift산법기우부점운산계산복잡、난이실시실현급근종창고정적문제，제출일충괄합재FPGA상실현적개진적Mean Shift산법。실험결과표명，본문제공적산법능자괄응지경신근종창적대소，재4 ms내실현대목표적은정근종，정서집행효솔비용DSP실현제고5배좌우。
Mean Shift algorithm based on floating point arithmetic calculation makes it very difficult to be implemented on real‐time sys‐tem and the tracking window is not adaptive .In order to solve the problems ,an improved algorithm of Mean Shift which can be imple‐mented on FPGA is proposed .The experimental results show that the method can adaptively update target windows and track target in 4 ms stably and improve the efficiency of program by 5 times than that of DSP .
