摘要
相关跟踪算法以其理论简单、直观、对图像质量要求不高等特点,一直是人们在跟踪算法方面的研究重点。针对相关跟踪抗遮挡能力差、容易受目标姿态变化和相似目标干扰的问题,使用了线性最小二乘估计对目标运动进行预测,同时采用高斯+1型函数和椭球面函数作为中心距离加权系数来改善跟踪模板使得模板更加突出中心位置信息。改进后的相关跟踪算法不但较好地解决了遮挡和相似目标干扰的问题,而且克服了局部遮挡和小范围形变对相关跟踪的影响。仿真试验表明:线性最小二乘估计及中心距离加权的改进型相关跟踪算法与传统的相关跟踪算法相比能更准确地跟踪目标,并具有一定的抗干扰能力。
Correlation tracking algorithm is always a focus of study because it is simple and intuitive in theory and has a moderate requirement to image quality. But the algorithm has bad anti-shielding ability and is easily affected by posture variation of the target and interfered by similar targets. To solve the problems, linear least-squares estimation was used in motion prediction, and ellipsoid function and Gauss ＋ 1 function were used as weighting coefficients of central distance, which made the central part of template more prominent. The improved correlation tracking algorithm is a better solution to the problems of the shielding and similar targets＇ interferences, and can overcome the effect of partial-shielding and small-scale deformation on tracking. Experimental results show that： compared with the traditional correlation tracking algorithm, the improved correlation tracking algorithm can track the target more accurately with betteranti-interference capability.
出处
《电光与控制》
北大核心
2011年第2期77-80,共4页
Electronics Optics & Control
基金
安徽省重点实验室基金(2007A0103013Y)
关键词
相关跟踪
最小二乘估计
运动预测
中心距离加权
椭球面
correlation tracking
Least-Squares Estimation （LSE）
motion prediction
weighted central-distance
ellipsoid
