Robust patch-based visual tracking using biologically inspired model
Ruyi Xu, Guoping Yan, Qing Jun Pan · 2012
We propose a novel algorithm for object tracking based on biologically inspired model (BIM) with hierarchical feedback architectures. The system integrates a support vector machine (SVM) classifier into a patch-based tracker. The template object is represented by a set of patches in the bottom layer of BIM. Each patch is tracked individually and a candidate position of object center is predicted based on the position of the corresponding patch in the current frame. Then the top layer feature of the candidate area is passed to the SVM classifier, and the top SVM score determines which patch prediction is reliable. Numerous experimental results validate that the proposed method is quite effective and robust to illumination changes, partial occlusions and slight appearance changes.