摘要
目前在轮船跟踪方面应用的算法存在明显的缺陷,比如遮挡、同色系干扰等,这些缺陷大大影响了跟踪的成功率,为此,在Camshift以及无损卡尔曼滤波的基础上提出一种跟踪算法。利用三帧差分提取运动特征,与色调、边缘特征进行融合,利用Camshift算法找到轮船的位置,利用无损卡尔曼滤波预测并校正跟踪框的中心和大小,利用Bhattacharyya距离描述跟踪情况。实验结果表明,在部分遮挡和同色系背景等复杂环境下,该算法能够准确跟踪到目标,鲁棒性和实时性较好。
Due to the problems that the existing tracking algorithm may lose the ship in complex situation such as partial occlusion and the same color background,a ship tracking algorithm was proposed based on Camshift with multi-feature and Kalman filter.Three frames were used to extract motion features and combine color and edge features.The location of ship was discovered using Camshift algorithm.The center and size of the tracking frame were predicted and corrected using the unscented Kalman filter,and Bhattacharyya distance was used to estimate the degree of occlusion.Experimental results show that the algorithm can track the target accurately under partial occlusion and the interference of same color,and it has better robustness and real-time performance.
作者
张永梅
刘萌萌
郭莎
ZHANG Yong-mei;LIU Meng-meng;GUO Sha(School of Computer Science,North China University of Technology,Beijing 100144,China;School of Electronic Information Engineering,North China University of Technology,Beijing 100144,China)
出处
《计算机工程与设计》
北大核心
2018年第2期479-485,共7页
Computer Engineering and Design
基金
国家自然科学基金项目(61371143)
北方工业大学基于内容感知的最优图像缩放技术研究与应用科研平台基金项目(XN054)
北方工业大学优势学科基金项目(XN078)
