Entity

Time filter

Source Type


Wang X.,Harbin Engineering University | Qiu C.,East China Research Institute of Photo electronic | Yin G.,Harbin Engineering University | Men Z.,Harbin Engineering University
IET Conference Publications | Year: 2012

A novel mean shift algorithm of v ideo images is proposed to track moving target in dynamic scene. The global compensation method is used to eliminate the background diversity caused by the motion of camera. Then the target is detected by difference algorithm and segmented by MRF method, so we can get the target template automatically. Finally mean shift is used to track moving target. This method solves the problem of orthodox mean shift tracking method which should select target region artificially. The experiments of the Coastguard standard video images and the practical with dynamic scene demonstrate that the proposed algorithm is effective to track moving target and it is precise and adaptive. Source


Wang X.,Harbin Engineering University | Yin G.,Harbin Engineering University | Men Z.,Harbin Engineering University | Qiu C.,East China Research Institute of Photo electronic
Proceedings - 2010 International Conference on Computational and Information Sciences, ICCIS 2010 | Year: 2010

Moving target extraction plays a prominent role in the whole target tracking process in image sequences with dynamic scene. This paper presents a new approach for moving target extraction. In order to automatically extract moving target and improve the accuracy, level set method is applied to extract moving target contour. Firstly, moving target is detected by difference algorithm. Secondly, Chan-Vese level set model is used to obtain the contour. The holes and discontinuous regions are filled through closing operation of mathematical morphology. According to the coordinates of the horizontal and perpendicular vertexes of binary image, the moving target is extracted. Finally, experiment results of the standard image sequences coastguard demonstrate that the proposed algorithm of this paper is highly adaptive and accurate. © 2010 IEEE. Source


Wang X.-M.,Harbin Engineering University | Yin G.-S.,Harbin Engineering University | Men Z.-G.,Harbin Engineering University | Qiu C.-G.,East China Research Institute of Photo electronic
Nanjing Li Gong Daxue Xuebao/Journal of Nanjing University of Science and Technology | Year: 2011

To exactly obtain global motion estimation in dynamic scene, this paper presents an adaptive global motion estimation method to eliminate outliers. The Best Bin First(BBF) method of the nearest neighbor search algorithm is used to match feature points extracted by the scale invariant feature transform(SIFT) algorithm. In order to improve the accuracy of feature matching, an improved Random Sample Consensus(RANSAC) algorithm is proposed that can eliminate outliers adaptively. The iterative number is controlled by the variance of motion magnitude of feature points. Through a camera motion model, accurate results of parameter estimation and background compensation are obtained. The proposed algorithm is tested by the Coastguard standard image sequence and the practical one with dynamic scenes. The experimental results are compared with the previous method, which demonstrates that the proposed algorithm is highly accurate and adaptive and that the speed is faster. Source


Wang X.,Harbin Engineering University | Yin G.,Harbin Engineering University | Men Z.,Harbin Engineering University | Qiu C.,East China Research Institute of Photo electronic
Harbin Gongcheng Daxue Xuebao/Journal of Harbin Engineering University | Year: 2011

In order to extract a moving target precisely in the dynamic scene, a novel moving target extraction algorithm based on Markov random fields (MRF) was proposed. The initial MRF parameters, composed of an isotropic clique of a double scale neighborhood MRF model, were automatically computed by the least square method according to the initial segmentation result. In order to obtain the MRF detection result, the iterated conditional mode (ICM) algorithm was used to estimate the maximum a posteriori (MAP). The holes and discontinuous regions were filled by closing operation of mathematical morphology. Finally, the moving target was extracted accurately according to the coordinates of the horizontal and perpendicular projection vertexes of a binary image. The experiments of the Coastguard standard image sequence along with a practical one with a dynamic scene demonstrate that the proposed algorithm is highly accurate, adaptive, and effective. Source

Discover hidden collaborations