許陳鑑Chen-Chien Hsu朱永青Yung-Ching Chu2019-09-032016-1-272019-09-032014http://etds.lib.ntnu.edu.tw/cgi-bin/gs32/gsweb.cgi?o=dstdcdr&s=id=%22GN060075002H%22.&%22.id.&http://rportal.lib.ntnu.edu.tw:80/handle/20.500.12235/95708本文提出了一種混合型即時影像多目標物體追蹤系統，結合粒子群聚最佳化法(Particle Swarm Optimization, PSO)與粒子濾波器(Particle Filter, PF)兩種演算法之優點，以提升多目標物體追蹤的性能。並且在可程式規劃系統晶片(System On a Programmable Chip, SOPC)之架構下，利用FPGA(Field Programmable Gate Array)的硬體電路，將此混合型演算法以軟硬體協同設計(HW/SW Co-design)之方式實現出來。此方式不僅可以達到硬體加速的功能，還能有彈性地設計整個電路，當軟體部分驗證完成後，即可以全硬體方式實現整個多目標物體追蹤系統，進一步提升整體系統的效能。This thesis presents a hybrid algorithm incorporating Particle Swarm Optimization (PSO) and Particle Filter (PF) for multiple-object tracking to improve the system performance. Based on the System on a Programmable Chip (SOPC) technique, we use hardware/software (HW/SW) co-design method to implement the hybrid algorithm on the FPGA circuit. As a result, the tracking efficiency can be greatly improved, while maintaining design flexibility for various applications. To further improve the performance of the multiple-object tracking system, full hardware implementation of the tracking system can be realized once the prototype testing of the system is completed.多目標物體追蹤可程式規劃系統晶片粒子群聚最佳化法粒子濾波器Multiple-object trackingSystem on a Programmable Chip (SOPC)Particle Swarm OptimizationParticle Filter