Multi: target Tracking of Double-threshold Region Segmentation Data Association Algorithm

The joint probability data association algorithm is widely used in the field of intelligent vehicle obstacle tracking,it is one of the effective ways to track multiple targets.But as the number of targets and echo increased,the confirm-matrix divisions expands exponentially to cause computational explosions,affecting the self-driving car to identify obstacles' requirements of real-time and accuracy.Therefore,proposed a double-threshold region segmentation JPDA algorithm.First, an ellipse and a position dynamic tracking gate are established according to the characteristics of the vehicle target,eliminate interference echo and reduce the number of joint events;Second, the filtered echo is clustered according to the area where the target is located;Finally, according to the different categories within the target and its dual threshold within the number of echoes were calculated for each type of association probability.The algorithm effectively reduces the number of joint events,and by the target segmentation can simplify the number of split confirmation matrix to reduce the amount of calculation.The actual experiment proves that the algorithm has real - time and validity.