以軟硬體協同設計之混合型即時影像物體追蹤系統

Hardware/Software Co-design of a Hybrid Object Tracking System Based on Particle Filter and Particle Swarm Optimization

林玟玲(Wen-Ling Lin)

指導教授 ： 易志孝

淡江大學/工學院/電機工程學系碩士班/碩士(2011年)

https://doi.org/10.6846/TKU.2011.00909

摘要

本文結合粒子濾波器(Particle Filter)與粒子群聚最佳化演算法(Particle Swarm Optimization, PSO)之優點，提出一種混合型即時影像物體追蹤系統，並於可程式規劃系統晶片(System on Program Chip, SOPC)之系統架構下，利用FPGA(Field Programming Gate Array)的硬體電路優勢，以軟硬體協同設計(HW/SW Co-design)之方式實現硬體加速之功能。作法上係利用所提出之切換機制，當粒子濾波器因物體移動速度太快而追丟時，便切換到PSO做一全域搜尋，而當PSO追蹤到目標物時，再切換到粒子濾波器做快速追蹤功能。並以多主從系統架構來設計硬體加速器，Nios II處理器(Nios II Processor)計算權重值，再以硬體電路進行粒子更新，藉由軟硬體緊密的合作，可以降低Nios II處理器的運算量，提升粒子濾波器與PSO演算法之執行效率，加快執行速度。也由於權重計算有彈性的設計方式，使得在解決各種問題時不需要重新設計硬體。實際結果顯示，利用SOPC軟硬體協同設計的技術所實現之影像物體追蹤系統可獲得良好之即時影像物體追蹤效果。

關鍵字

粒子濾波器 ； 粒子群聚最佳化法 ； 目標物追蹤 ； SOPC ； FPGA ； 軟硬體協同設計

並列摘要

This paper presents a hardware/software co-design method for implenting a hybrid object tracking system based on particle filter and Particle Swarm Optimization via System on Program Chip (SOPC) technique. Practice on the system using the proposed switching method When the particle filter lost the tracking because object moving too fast,it will switch to PSO to do a global search. When the PSO to tracking the object, it will switch to the particle filter to do fast tracking. Considering both the execution speed and design flexibility, we use a NIOS II processor to calculate weight for each particle and a hardware accelerator to update particles. As a result, execution efficiency of the proposed hardware/software co-design method of particle filter and Particle Swarm Optimization is significantly improved while maintaining design flexibility for various applications. To demonstrate the performance of the proposed approach, a real-time object tracking system is established and presented in this paper. Experimental results have demonstrated the proposed method have satisfactory results in real-time tracking of objects in video sequences.

並列關鍵字

Particle Filter ； Particle Swarm Optimization ； Object Tracking ； SOPC ； FPGA ； HW/SW co-design

參考文獻

[17] 王瑞斌，以軟硬體協同設計之粒子濾波器及其在影像追蹤之應用，淡江大學電機工程學系碩士班碩士論文(指導教授：許陳鑑)，2010。

[22] 鄭明育，演化式物體影像追蹤與傾斜定位，淡江大學電機工程學系碩士班碩士論文(指導教授：許陳鑑)，2009。

[1] Y. C. Ho and R. C. K. Lee, “A Bayesian approach to problems in stochastic estimation and control,” IEEE Transactions on Automatic Control, vol. AC-9, pp. 333-339, 1964.

[2] J. H. Kotecha and P. M. Djuric, “Gaussian Particle Filtering,” IEEE Transactions on Signal Processing, vol. 51, no. 10, pp. 2592-2601, 2003.

[3] N. de Freitas, “Rao-Blackwellised particle filtering for fault diagnosis,” IEEE Aerospace Conference Proceedings, vol. 4, pp. 4-1767-4-1772, 2002.