软硬件协同设计算法的嵌入式人脸识别系统

摘要

研究人脸识别系统的软硬件设计优化问题,软硬件协同设计是一个多约束条件、多目标的组合优化问题,单一方法难以找到最优协同设计方案.为此,提出一种采用遗传-蚁群算法的嵌入式人脸识别系统软硬件协同设计方法.首先建立软硬件协同设计问题的数学模型,然后利用遗传算法找到问题的可行解,最后采用蚁群算法在可行解中找到全局最优协同设计方案.仿真结果表明,遗传-蚁群算法很好的利用了两种算法优势,解决了单一算法存在局部极优、早熟等难题,提高了软硬件协同设计方案的求解效率,可以找到更优的嵌入式人脸识别系统软硬件协同设计方案.%Software/hardware collaborative design problems of human face recognition embedded system was studied.In order to get the optimal performance of the embedded system,this paper put forward a design algorithm based two phase algorithm.Firstly,the mathematical model of hardware software collaborative design of embedded system design the problem was established,and then a genetic algorithm was adopted to find the feasible solution.Finally,the ant colony algorithm was adopted to find global optimal design in the feasible solution.The simulation results show that this proposed algorithm solves the problem in single heuristic algorithm to improve the solving efficiency,and it can find the optimal schedule for hardware software collaborative design of embedded system design and has a certain practical value.