A Review of Point Target and Extended Target Tracking Algorithms

In this paper, we investigate the methods of multitarget tracking in imaging process, focusing on both point targets and extended targets. Point target imaging is characterized by low signal-to-noise ratio (SNR), small area, and indistinct appearance features. The study primarily introduces Detection Before Tracking (DBT) and Tracking Before Detection (TBD) algorithms, including Joint Probabilistic Data Association (JPDA), Multiple Hypothesis Tracking (MHT), and Cardinalized Probability Hypothesis Density (CPHD) methods for data association, and analyzes the principles and application scenarios of TBD methods such as Hough transform, dynamic programming, and particle filtering. Regarding the problem of multi-extended target tracking, the study emphasizes the utilization of the Harris corner detection principle for feature extraction, the random hyper-surface model for modeling extended targets, and the trajectory establishment problem based on the Probability Hypothesis Density (PHD) algorithm.