An Adaptive KCF Tracking Via Multi-feature Fusion

Abstract

To tackle the problem of target scale changed, too monotonous target feature, or track cumulative errors in Kernelized correlation filters(KCF), the paper proposes a self-adaptive KCF tracking algorithm employed multi-feature fusion. KCF tracking algorithm is improved based on location prediction, multi-feature fusion and bilinear interpolation. Among them, to facilitate better representation of the target's appearance model, make target tracking more robust, the multi-feature fusion is integrated (Hue Saturation Value, HSV) color features, grayscale features and improved (Histogram of Oriented Gradient, HOG) features. Both qualitative and quantitative evaluations on some object tracking benchmark show that the proposed tracking method achieves superior performance compared with other state-of-the-art methods.