用于指尖定位的多目标分布估计算法

A Multi-objective Estimation of Distribution Algorithm for Fingertip Localization

为了高效地求解指尖定位优化模型,进而准确并且快速地得到指尖位置,对用于求解该优化模型的分布估计算法进行研究.通过对这一算法的分析可知,由于指尖定位优化模型的最优解分量分布在指尖中心的附近,采用的算法参数应该符合上述分布规律.实验结果表明,对于上述算法中的决策变量维数、种群规模、最大采样方差、最小采样方差,它们均取最佳值时,得到的计算结果优于已有方法.因此,上述4个参数均取最佳值时,能够准确并且快速地得到指尖位置.

In order to solve the optimization model for fingertip localization efficiently,and obtain the fingertip positions accurately and rapidly,the estimation of distribution algorithm for the above optimization model was explpreg in this paper.The analytical results of this algorithm showed that due to the optimal solution components of this optimization model distribute around the fingertip centers,the used algorithm parameters should accord with the above distribution law.The experimental results showed that when the values of the decision variable dimension,population size,maximum sampling variance,and minimum sampling variance of this algorithm were their best values.The results obtained by the proposed method were better than those by the existing methods.Thus,when the values of the above four parameters were their best values,the fingertip positions could be obtained accurately and rapidly.