一种基于新型遗传算法的块运动估计算法

提出了一种基于新型遗传算法的块匹配运动估计算法。该算法把块运动向量作为遗传染色体,经过选择变异等操作,将随机搜索与特定目标搜索相结合,解决了以往快速搜索算法易陷于局部最优的问题,同时该算法中所带的模糊评价函数使得对个体的评价更合理、客观,该算法还将运动矢量空间偏置特性用于初始种群的选取,进一步提高了算法性能。实验结果表明,该算法性能上接近于FSA,速度却接近于TSS。

A Block Motion Estimation Method Based on a Novel Genetic Algorithm

This article proposes a block matching motion estimation algorithm based on a novel genetic algorithm.This method solves the problem of being trapped to local optimal solution that most fast block matching motion estimation algorithms can cause,by treating motion vectors as chromosome and putting genetic operation on it,In addition,the fuzzy fitness function it takes can evaluate the fitness of individual more objectively,To make this method more effectively,we select the initial population based on the coherence rule within the motion vector space,Experimental results show that this method has the performance close to FSA,but has the speed close to TSS.