基于贪心退火遗传算法的车辆图像分割研究  

Traffic Image's Threshold Value Division Based on Greedy Annealing Genetic Algorithm

作　　者：汤晖[1] 李润 许伦辉[1] 黄艳国[1] 

机构地区：[1]江西理工大学,赣州341000 [2]广州市交通管理科学研究所,广州510000

出　　处：《交通与计算机》2007年第4期19-21,25,共4页Computer and Communications

基　　金：国家自然科学基金项目(批准号:60664001);江西省自然科学基金项目(批准号:0511030)资助

摘　　要：针对交通流实时检测系统图像分割的阈值自动优化选取问题,通过将贪心思想和模拟退火思想引入到遗传算法中,设计了贪心退火遗传算法(GAGA),以最大类方差函数作为遗传算法中适应度的评价函数,利用贪心算法局部搜索能力强和退火遗传算法全局搜索能力强的特点,实现图像阈值的自动优化选取。M ATLAB仿真数据表明,贪心退火遗传算法较基本遗传算法具有更强的寻优性能。将其应用于交通流检测系统,可增强整个系统的实时性和鲁棒性。In view of the threshold value automatic optimization selection of traffic flow＇s image division in real-time examination system, the thoughts of greedy and simulation of annealing were introduced into genetic algorithm. A greedy annealing genetic algorithm（GAGA）was designed. With the Otsu function being used as the fitness evaluation function, and by using the characteristic of greedy algorithm＇s strong ability of partial search and annealing genetic algorithm＇s strong ability of overall search, threshold value automatic optimization selection of traffic flow＇s image was realized. The MATLAB simulation data indicate that this algorithm has stronger optimal ability and faster convergence speed than the basic genetic algorithm. Its application to the detection system of traffic flow can strengthen the timeliness and robustness of the system.

关 键 词：贪心退火遗传算法 阈值分割 交通流 实时性 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]