运动目标检测是计算机视觉领域极具挑战性的难题,针对CP3(Co-occurrence Probability based Pixel Pairs)算法的计算量大、实时性差,运动目标遮挡检测不完全等问题提出了改进方法:一种融合SLIC(Simple Linear Iterative Cluster)超像素...运动目标检测是计算机视觉领域极具挑战性的难题,针对CP3(Co-occurrence Probability based Pixel Pairs)算法的计算量大、实时性差,运动目标遮挡检测不完全等问题提出了改进方法:一种融合SLIC(Simple Linear Iterative Cluster)超像素与CP3算法的目标检测算法(Co-occurrence Probability based Super Pixel Pairs)。该算法首先采用SLIC算法将当前帧分割为超像素集合,并对线性相关超像素对的亮度增量差进行了单高斯建模,构建线性相关背景模型。实验结果表明,改进算法大幅缩减了建模时间,使运行速度提高了3倍,对遮挡也有较好的鲁棒性,使目标检测综合指标F-measure提高了5.3%。展开更多
This thesis offers the general concept of coefficient of partial correlation.Starting with regres-sion analysis,the paper,by using samples,infers the general formula of expressing coefficient of partial correlation by...This thesis offers the general concept of coefficient of partial correlation.Starting with regres-sion analysis,the paper,by using samples,infers the general formula of expressing coefficient of partial correlation by way of simple correlation coefficient.展开更多
文摘运动目标检测是计算机视觉领域极具挑战性的难题,针对CP3(Co-occurrence Probability based Pixel Pairs)算法的计算量大、实时性差,运动目标遮挡检测不完全等问题提出了改进方法:一种融合SLIC(Simple Linear Iterative Cluster)超像素与CP3算法的目标检测算法(Co-occurrence Probability based Super Pixel Pairs)。该算法首先采用SLIC算法将当前帧分割为超像素集合,并对线性相关超像素对的亮度增量差进行了单高斯建模,构建线性相关背景模型。实验结果表明,改进算法大幅缩减了建模时间,使运行速度提高了3倍,对遮挡也有较好的鲁棒性,使目标检测综合指标F-measure提高了5.3%。
文摘This thesis offers the general concept of coefficient of partial correlation.Starting with regres-sion analysis,the paper,by using samples,infers the general formula of expressing coefficient of partial correlation by way of simple correlation coefficient.
