Based on significant deformation of the object tracking method and system

The present invention discloses a deformation object tracking algorithm and a system based on visual saliency, the method comprising the steps of: S1: artificial calibration target area; S2: by an image segmentation algorithm based on visual saliency acquires partial comprising foreground object is significantly of Atlas; S3: dividing the foreground area; S4: establishing a target foreground model; S5: initializing the model of the foreground color histogram based on color characteristics of the tracker; S6: an optical flow tracker feature points based on candidate target acquisition and its determination; S7: acquiring a target final position; S8: learning objectives and updates. While the invention this be significant processing also contemplated computational efficiency of the algorithm, and automatically segmented foreground information by the method of local window size to establish a softened model by foreground information matrix and generating a color histogram corresponding to the model, a histogram Figure match, increasing the ability to judge the tracker, while the learner through continuous learning new model targets.