Fig. 1. Framework of long-term tracking with CFMK

Fig. 2. Response fusion of multiple features

Fig. 3. Tracking results for nine kinds of trackers on different databases. (a) Tracking success rate and (b) tracking precisionfor OTB-13 database; (c) tracking success rate and (d) tracking precision for OTB-15 database

Fig. 4. Comparison of tracking success rate for each algorithm under eleven kinds of scenes. (a) In-plane rotation; (b) low resolution; (c) object occlusion; (d) out of view; (e) out-of-plane rotation; (f) scale variation; (g) fast motion; (h) background clutter; (i) motion blur; (j) object deformation; (k) illumination variation

Fig. 5. Tracking result graphs in partial sequences for six kinds of tracking algorithms. (a) BlurOwl; (b) Liquor; (c) Box; (d) Jogging 1; (e) Jogging 2; (f) Tigger

Table 1. Mean DP and OP on different databases for proposed algorithm under different values of K

Table 2. Mean tracking performance parameters on different databases for nine kinds of tracking algorithms