摘 要: 針對人工蜂群算法存在的收斂速度較慢,易陷入局部最優解的問題,提出一種改進的人工蜂群優化算法,并應用于數字圖像相關的整像素位移搜索中。該算法借助相關度值的變化來動態調整跟隨蜂的搜索步長,平衡其全局和局部的搜索能力;偵察蜂利用遺傳算法的交叉運算產生新解,改善全局搜索能力。實驗結果表明,改進的算法能有效地提高收斂速度,改善整像素位移搜索的性能。
關鍵詞: 人工蜂群算法; 數字圖像相關; 變形; 整像素位移; 動態步長; 交叉運算
中圖分類號: TN919?34; TP301.6 文獻標識碼: A 文章編號: 1004?373X(2013)24?0070?03
Application of improved artificial bee colony algorithm in digital image correlation
YANG Song1,2, SHAO Long?tan1, JIN Feng3, SHI Peng?hui4, CHI Jian?wei5
(1. State Key Laboratory of Industrial Equipments, Dalian University of Technology, Dalian 116024, China;
2. Education Technology Computing Center, Dalian Ocean University, Dalian 116023, China;
3. Dalian Municipal Meteorological Bureau, Dalian 116001, China; 4. College of Information Engineering, Dalian Ocean University, Dalian 116023, China;
5. College of Science, Dalian Ocean University, Dalian 116023, China)
Abstract: Aiming at the problems of slow convergence speed and easy to fall into local optimal solution, which exist in artificial bee colony algorithms, an improved artificial bee colony optimization algorithm is proposed for the pixel displacement search in digital image correlation. In the algorithm, the search step of the following bee is adjusted dynamically according to the changes of the correlation to balance the capability of global search and local search, and the scout bee employs crossover operation of genetic algorithm to generate new solution for improving the global search capacity. The experimental results show that the improved algorithm can enhance the convergence capability effectively and improve the performance of integer pixel displacement search.
Keywords: artificial bee colony algorithm; digital image correlation; deformation;pixel displacement; dynamic step; crossover operarion
數字圖像相關方法(Digital Image Correlation Method,DICM),又稱為數字散斑相關方法 (Digital Speckle Correlation Method,DSCM) 是現代光學測量技術的重要方法之一[1],用于實現物體表面的位移和變形的測量。數字圖像相關方法在1980年由Peters,Ranson和Yamaguchi同時獨立提出的。該方法不僅具有其他光學測量方法所具有的全場、非接觸測量等優點外,還具有光路簡單、對測量環境要求低、便于開展工程現場測量等優點[2],是一種很有應用前途的光學測量方法。
數字圖像相關法研究的核心內容是相關搜索方法。早期采用逐點搜索法,這會花費大量時間,后來很多學者對其進行改進,相繼提出了Newton?Rapson算法[3]、爬山法[4]、十字搜索法[5]、小波變換法[6]和神經網絡法[7]等,這些算法嚴重依賴于初值的選取,且部分在理解和實現上比較復雜。相比之下,群智算法有諸多優點,可直接把目標函數值作為搜索信息,避免函數求導,可解決目標函數較復雜的問題。……