基于视觉感知的运动目标跟踪算法

摘要

针对运动目标跟踪问题,为解决跟踪过程中因遮挡、目标尺度变化等易造成跟踪失败的现象,提出一种基于视觉感知的跟踪算法.该算法以神经元响应为视觉特征,首先从自然图像中学习初级视皮层细胞感受野；然后计算背景图像和视频序列图像的神经元响应并得出差值,与动态阈值比较,识别出运动目标,通过迭代实现目标跟踪.多类别实验结果表明,该算法实现了运动目标稳定跟踪,目标跟踪准确率达93.5％且鲁棒性增强,与典型算法Camshift和SIFT相比,提高了跟踪算法的准确性和鲁棒性.%Abtract:For moving object tracking problems,this paper put forward an object algorithm based on visual perception,in order to solve the tracking failure phenomenon which was caused by the occlusion of tracking process and the change of object scale.This algorithm used neural response as visual features.First of all,it studied primary visual cortex cell receptive field from the natural image and calculated the neurons response of background images and video sequences,and drew the difference.Then it identified the moving object by comparing with the dynamic threshold value.Finally,it achieved object tracking by iteratively.The multi-category experiment results show that the algorithm achieves tracking of the moving object stability,and that the object tracking's accuracy rate has achieved 93.5％ and its robustness is enhanced.Compared with the typical algorithm,such as the Camshift and the SIFT,this algorithm improves the accuracy and robustness of the tracking algorithm.