摘要
利用七阶不变矩对图像进行相关匹配,可以有效抑制几何失真的影响;在匹配的过程中对输入图像进行3层金字塔分解,在每一层上利用高效、并行的遗传算法进行匹配,搜索最优匹配点;实验结果表明该算法能够实时计算运动目标的位置,并且在目标发生缩放、旋转、光照等变化时仍能够长时间稳定跟踪目标。
A fast algorithm of correlation tracking based on seven moment invariants is proposed. The input image is decomposed to multiresolution image with pyramid structure of three layer. In every layer, genetic algorithm is used to find the optimal matching point so as to reduce the computation cost. The experimental results indicate that this method can locate the target accurately and fast , can track the target stably in a long time while the target rotating or the size of target and the intensity of sunlight rays changing.
出处
《计算机测量与控制》
CSCD
2005年第9期995-996,999,共3页
Computer Measurement &Control
关键词
目标跟踪
特征匹配
遗传算法
不变矩
target tracking
feature based matching
genetic algorithm
moment invariants
