Effective moving object tracking using modified flower pollination algorithm for visible image sequences under complicated background

In real time moving object tracking, various unconstraint factors such as severe occlusion and background clutter remain a challenge for developing an effective yet robust tracking method. In this study, a novel moving object tracking algorithm based on the modified flower pollination algorithm (MFPA) is proposed. A search window with the features of centroid coordinates and width of the search window is used to locate the position of the moving object. The sub-image within the search window is then extracted. The hue, saturation and value (HSV) histogram of the extracted region is calculated in order to model the object appearance representation. Subsequently, the Bhattacharyya distance between the HSV histograms from two consecutive frames is formulated as the fitness function of MFPA, in which maximizing the similarity of both histograms is sought after. The comparative experimental results show that the developed algorithm outperforms others in terms of efficiency and accuracy. •We propose a novel modified flower pollination algorithm for moving object tracking.•Its effectiveness is tested in a series of benchmark videos.•Performance comparison with other algorithms is made.•The proposed moving object tracking system shows better performance than the others.