Sequential Particle Filter For Multiple Object Tracking
Nam Trung Pham, Karianto Leman, Teck Wee Chua · 2011
Tracking objects in a scene under different degree of occlusions is still a challenge in computer vision. In this paper, we aim to tackle this problem with a jump observation model that includes both visible objects and occluded objects in a sequential particle filter tracking technique. The technique is extended from the sequential estimation method. An hierarchical structure is introduced to model visible objects and its objects that are occluded. With this structure, we are able to handle the weak measurement observation of the occluded objects and try to track objects under occlusions. 1