Algorithm for Target tracking using modified image filtering based on statistical features and GMM image modeling

This paper presents an algorithm of target tracking using image modeling based on Gaussian Mixture Model (GMM) and modified filtering based on statistical feature information designed to improve the accuracy and robustness of the target tracking algorithm in video images for surveillance. The algorithm extracts the pixels of the foreground image including noise through gaussian mixture modeling, and designs template based on the structural element idea of morphological operation to count the sum of pixel values matching the template, then selectively filter out image noise according to the threshold. The experimental results conclude that the algorithm in this paper has good noise filtering performance as well as accuracy of target tracking and effectively retaining target information.