基于Hausdorff距离与混沌神经网络的图像匹配.pdfVIP

基于Hausdorff距离与混沌神经网络的图像匹配 石争浩1’2，黄士坦2柳虹3 710054；3浙江工商大学计算机与信息工程学院，杭州310035) 摘要：由于传统的图像匹配方法受光照、噪声和景物遮挡等的影响比较大，并且需要建立匹配图像对之间的对应关 系，因而使实际图像的匹配变得困难。为了克服上述缺陷，以便快速地进行图像匹配，引入Hausdorff距离构造匹配 相似度量和混沌优化的Hopfield神经网络的图像匹配算法。实验表明，该算法较之常规图像匹配算法具有更快的匹配 速度和更高的匹配概率。 关键词：图像匹配；Hausdorff距离；混沌 BasedonHausdorffDistanceandChaoticNeural ImageMatching Network SHI Shi—tan2，LIU Zheng．ha01～，HUANGHon93 (1Schoolof ComputerScienceEngineenng，Xi’anUniversityofTechnology，Xi’an Xi’allInstituteof of Micro-electronics，Xi’an710054，School ComputerInformationEngineering，ZheJiangGongShang 3 University，Hangzhou10035) Abstract：A basedon neuralnetworkis the of discussed，where matching imagematchingalgorithmHopfield problemimage istreatedastheminimizationofthe functionofthe neuralnetwork．An neuralnetwork energy Hopfield optimizedHopfield basedontheHausdorffdistanceandChaos is toovercomethelimitationof imagematchingalgorithm theorypresented neural as tolocal to motionandtransformation．TheHausdorff Hopfieldnetwork，suchtending minima，sensitiveobject distanceisusedtomeasurethe oftwo isusedto thesearch of Neural similarityobjects，Chaosoptimize processHopfield anew formulationfor invariant is resultsdemonstratethe Network，andenergy general matchingderived．Experimental andtheeffectivenessofthe method． efficiency proposed Distance，Chaos Keywords：ImageMatching，Hausdorff