Object Tracking Based On Feature Manifold And Self-learning Method

Object tracking has always been a hot topic in Computer Vision. In Object tracking field, local feature based object tracking approaches have been promising in solving the tracking problems such as occlusions and illumination variations. However, existing approaches typically model feature variations using prototypes, and this discrete representation can not capture the gradual changing property of local appearance. In this paper, we propose to model each local feature as a feature manifold to characterize the smooth changing behavior of the feature descriptor. The manifold is constructed from a series of transformed images simulating possible variations of the feature being tracked. We propose to build a collection of linear subspaces which approximate the original manifold as a low dimensional representation. This representation is used for object tracking. Object location is located by a feature-to-manifold matching process.In order to handle scale change dynamically in tracking process, our method will fit a Homography Matrix according to feature-to-manifold matching information. With this matrix, we are able to transform the coordinates of object position in current frame to obtain the object scale and location information in next frame.What’s more, our tracking method can update the manifold status, add new feature manifolds and remove expiring ones adaptively according to object appearance. We show both qualitatively and quantitatively this method significantly improves the tracking performance under occlusions and appearance variations using standard tracking dataset.