An Automatic Track Initialization Based On Combining The Hough Transform Method And A Multi-level Filter

Abstract

Modern surveillance systems require high accuracy in track initialization. However, data processing in dense clutter environments is always a challenging issue as the majority of the observations from radars are false. To solve this problem, we propose to use a method that combines the Hough transform technique with a multi-level filter system. This strategy took advantage of the advantages of the Hough Transform algorithm while overcoming its limits by utilizing the strength of the intuitive method via orbital information filters. This method has proven effective in the data initialization procedure, particularly in densely cluttered environments, it can perform track initialization fast while eliminating the appearance of false targets. We have conducted different experiments on the same data set with the same configuration parameters within 10 radar scans. We got 10 trajectories without any false targets using the multi-level filter system, Without it, we obtained 145 trajectories, of which 135 trajectories are false ones.