Fig. 1. Flow chart of overall algorithm

Fig. 2. Dictionaries learned through K-SVD algorithm for three patch sizes

Fig. 3. Process of computing HSC

Fig. 4. Precision plot of different algorithms tested on 20 sequences

Fig. 5. Success rate plot of different algorithms tested on 20 sequences

Fig. 6. Precision plots of all attributions. (a) Illumination variation; (b) out-of-plane rotation; (c) scale variation; (d) deformation; (e) occlusion; (f) motion blur; (g) fast motion; (h) in-plane rotation; (i) out of view; (j) background clutter; (k) low resolution

Fig. 7. Success rate plots of all attributions. (a) Illumination variation; (b) out-of-plane rotation; (c) occlusion; (d) scale variation; (e) deformation; (f) fast motion; (g) motion blur; (h) in-plane rotation; (i) background clutter; (j) out of view; (k) low resolution

Fig. 8. Tracking results of the proposed algorithm and another nine trackers on nine image sequences. (a) Out-of-plane rotation (birds and running_rhino sequences); (b) deformation (birds and crouching sequences); (c) background clutter (crowd and mixed_distractors sequences); (d) occlusion (street and hiding sequences); (d) scale variation (jacket and selma sequences)