机构地区: 四川大学电子信息学院
出 处: 《四川大学学报(自然科学版)》 2005年第1期82-86,共5页
摘 要: 心肌细胞的搏动是一种弹性运动,通过对搏动过程中特征点的运动矢量检测,可以莸得细胞收缩的生物力学特性.对数搜索、三步搜索等快速搜索算法都依赖于匹配误差单调减小这一假设,而这种假设在许多情况是不真实的,这就导致图象匹配仅仅收敛于某些局部优化值.在经典遗传算法的基础上,作者提出了自适应交叉算子和变异算子,应用于柔性运动矢量检测.解决了心肌细胞序列图象特征点的运动矢量检测问题. Beating of crdiac myocyte is elastic motion.The biomechanics characters of myocyte can be ob-tained through the analysis of theparticular points'movements.The algorithms such as logarithm-search and three-step-search are rely on an assumption that the matching error decreases monotnically as the searched point moves closer to the global potimum.But this assumption is far from reality and the algorithms may con-vere only to a local optimum.Genetic Algorithm(GA)is a kind of global optimum algorithm.An adaptive crossing operator and an adaptive mutating operator are proposed based on the classic GA.Using these opera-tors motion vector of the beating of cardiacmyocyte was detected.
领 域: [生物学]