基于混合蛙跳优化算法的图像畸变校正研究  

Image distortion correction based on shuffled frog leaping algorithm

作　　者：刘超 秦川[2] 陶忠[2] 王婉丽[2] 骆媛[2] 王晶[2] Liu Chao Qin Chuan Tao Zhong Wang Wanli Luo Yuan Wang Jing(Military Representative Office of Army Aviation Bureau in Xi＇an, Xi＇an 710065, China Xi＇an Institute of Applied Optics, Xi＇an 710065, China)

机构地区：[1]陆军航空兵军代局驻西安地区军代室,陕西西安710065 [2]西安应用光学研究所,陕西西安710065

出　　处：《应用光学》2017年第2期243-249,共7页Journal of Applied Optics

基　　金：国家国际科技合作专项(2015DFA10670)

摘　　要：图像畸变校正是分布式孔径系统(DAS)需要解决的先期问题。介绍了描述图像径向与切向畸变的Brown模型及其畸变量(适应度)度量标准;讨论了智能优化算法之一的混合蛙跳算法(SFLA)及其初始种群拉丁超立方抽样(LHS)算法;分别对光线追迹畸变模型、Brown畸变模型进行了图像仿真和畸变校正处理。结果显示,原始点与校正点的最大误差距离在2个像素以内,验证了算法的有效性。最后应用该方法到实际广角CCD相机拍摄的靶纸校正中,得到了较满意的结果。该算法和仿真分析结果对DAS系统的研制具有一定理论和实验意义。Image distortion correction is a prerequisite for distributed aperture system （DAS）. Brown model and its aberration （fitness） metric describing radial and tangential distortions of image are introduced. As one of intelligent optimization algorithms, shuffled frog leaping algorithm （SFLA） and its initial population Latin hypercube sampling （LHS）,are discussed. Image aberration model （from ray tracing method） and Brown distortion model are simulated and corrected respectively. Results show that characteristic point’s positions between origional and corrected are within 2 pixels, which verifies effectiveness of algorithm. Finally, this correction method is applied to correction of target paper by wideangle CCD camera, and satisfactory results are obtained. The algorithms and simulation results have certain theoretical and experimental significance to development of DAS equipment.

关 键 词：广角畸变 蛙跳算法 拉丁超立方抽样 图像校正 Brown系数 

分 类 号：TN202[电子电信—物理电子学] TP751.1[自动化与计算机技术—检测技术与自动化装置]