Directed at the problem of occlusion in target tracking,a new improved algorithm based on the Meanshift algorithm and Kalman filter is proposed.The algorithm effectively combines the Meanshift algorithm with the Kalma...Directed at the problem of occlusion in target tracking,a new improved algorithm based on the Meanshift algorithm and Kalman filter is proposed.The algorithm effectively combines the Meanshift algorithm with the Kalman filtering algorithm to determine the position of the target centroid and subsequently adjust the current search window adaptively according to the target centroid position and the previous frame search window boundary.The derived search window is more closely matched to the location of the target,which improves the accuracy and reliability of tracking.The environmental influence and other influencing factors on the algorithm are also reduced.Through comparison and analysis of the experiments,the modified algorithm demonstrates good stability and adaptability,and can effectively solve the problem of large area occlusion and similar interference.展开更多
结合Haar型特性局部二元模式(HLBP)的图像纹理特征提取方法,提出一种新的目标跟踪算法,并将其运用到Meanshift框架中。将Visual Studio 2010和opencv2.4.9作为实验平台,将所提算法的实验结果与传统Meanshift跟踪算法、基于局部二元模式(...结合Haar型特性局部二元模式(HLBP)的图像纹理特征提取方法,提出一种新的目标跟踪算法,并将其运用到Meanshift框架中。将Visual Studio 2010和opencv2.4.9作为实验平台,将所提算法的实验结果与传统Meanshift跟踪算法、基于局部二元模式(LBP)纹理特征的Meanshift跟踪算法进行对比分析。实验结果表明,所提算法在背景复杂或背景简单的情况下都表现出了稳健而准确的跟踪特性,且在部分遮挡的情况下仍可以正确地跟踪目标。展开更多
基金Supported by the Scholarship of China Scholarship Council(CSC)(201606935043)
文摘Directed at the problem of occlusion in target tracking,a new improved algorithm based on the Meanshift algorithm and Kalman filter is proposed.The algorithm effectively combines the Meanshift algorithm with the Kalman filtering algorithm to determine the position of the target centroid and subsequently adjust the current search window adaptively according to the target centroid position and the previous frame search window boundary.The derived search window is more closely matched to the location of the target,which improves the accuracy and reliability of tracking.The environmental influence and other influencing factors on the algorithm are also reduced.Through comparison and analysis of the experiments,the modified algorithm demonstrates good stability and adaptability,and can effectively solve the problem of large area occlusion and similar interference.
文摘为满足车辆行驶时能对各种车道线(实线、虚线、直道、大弯道)准确识别,提出一种基于Meanshift原理和RANSAC(Random Sample Consensus)算法的车道识别方法;该方法首先利用改进的最大熵阈值分割方法和图像灰度概率密度特征对左右车道线目标进行初定位,动态地建立车道线ROI(Region of Interests),然后运用Meanshift算法对左右车道线进行精确定位,最后利用RANSAC算法对各搜索框中候选车道线的重心进行筛选,并采用最小二乘法对左右车道线进行拟合;实验结果表明,该方法可以识别各种车道线型,并具有较好的鲁棒性;车道检测平均时间为80ms/f,车道跟踪平均时间为40ms/f。
文摘结合Haar型特性局部二元模式(HLBP)的图像纹理特征提取方法,提出一种新的目标跟踪算法,并将其运用到Meanshift框架中。将Visual Studio 2010和opencv2.4.9作为实验平台,将所提算法的实验结果与传统Meanshift跟踪算法、基于局部二元模式(LBP)纹理特征的Meanshift跟踪算法进行对比分析。实验结果表明,所提算法在背景复杂或背景简单的情况下都表现出了稳健而准确的跟踪特性,且在部分遮挡的情况下仍可以正确地跟踪目标。