Object tracking in video by egg shape boundary model and properties of HSV colour space

Abstract

In this paper, we propose an egg shape boundary model for object tracking in video sequence. The target in the current frame is bounded initially and the features of the target object are extracted. Each pixel in the boundary is transformed either as grey colour or as true colour in the HSV colour space and is used for representing the target pixel inside the boundary. In this model, the acceleration and velocity of the moving target object is considered for forming the egg shape boundary in the next frame by calculating the tangent points. A moving predicted centre technique is proposed for failure detection and recovery. The proposed method is suitable for tracking fast moving target object. For evaluating the performance, we have compared the results with other recently proposed methods in PETS video dataset and the results are found to be encouraging. In addition, it has been noticed that the egg shape boundary model takes less time compared to other model for processing each frame to form the target boundary.