Fast Scale Adaptive Kernel Correlation Filtering Algorithm for Target Tracking

Xuedong He; Shengzong Zhou

doi:10.3788/LOP55.121501

Journals >Laser & Optoelectronics Progress >Volume 55 >Issue 12 >Page 121501 > Article

Laser & Optoelectronics Progress
Vol. 55, Issue 12, 121501 (2018)

Fast Scale Adaptive Kernel Correlation Filtering Algorithm for Target Tracking

Xuedong He and Shengzong Zhou^*

Author Affiliations

Fujian Institute of Research on the Structure, Chinese Academy of Sciences, Fuzhou, Fujian 350002, China

show less

DOI: 10.3788/LOP55.121501 Cite this Article Set citation alerts

Xuedong He, Shengzong Zhou. Fast Scale Adaptive Kernel Correlation Filtering Algorithm for Target Tracking[J]. Laser & Optoelectronics Progress, 2018, 55(12): 121501 Copy Citation Text

show less

Precision and success rate of 26 sequences. (a) Precision of OPE; (b) precision of SRE; (c) precision of TRE; (d) success rate of OPE; (e) success rate of SRE; (f) success rate of TRE

Fig. 1. Precision and success rate of 26 sequences. (a) Precision of OPE; (b) precision of SRE; (c) precision of TRE; (d) success rate of OPE; (e) success rate of SRE; (f) success rate of TRE

Download full size

Tracking effect of KCF and FSAKCF on sequences of David, CarScale, and Dog1. (a) David, tracking frame number of 11; (b) David, tracking frame number of 149; (c) David, tracking frame number of 462; (d) CarScale, tracking frame number of 11; (e) CarScale, tracking frame number of 167; (f) CarScale, tracking frame number of 242; (g) Dog1, tracking frame number of 11; (h) Dog1, tracking frame number of 733; (i) Dog1, tracking frame number of 1290

Fig. 2. Tracking effect of KCF and FSAKCF on sequences of David, CarScale, and Dog1. (a) David, tracking frame number of 11; (b) David, tracking frame number of 149; (c) David, tracking frame number of 462; (d) CarScale, tracking frame number of 11; (e) CarScale, tracking frame number of 167; (f) CarScale, tracking frame number of 242; (g) Dog1, tracking frame number of 11; (h) Dog1, tracking frame number of 733; (i) Dog1, tracking frame number of 1290

Download full size

Sequence	Framenumber	Attribute
Boy	602	SV, MB, FM, IPR, OPR
Box	1161	IV, SV, OCC, MB, IPR,OPR, OV, BC, LR
David	471	IV, SV, OCC, DEF,MB, IPR, OPR
Car4	659	IV, SV
ClifBar	472	SV, OCC, MB, FM,IPR, OV, BC
CarScale	252	SV, OCC, FM, IPR, OPR
Crossing	120	SV, DEF, FM, OPR, BC
Dancer	225	SV, DEF, IPR, OPR
Doll	3872	IV, SV, OCC, IPR, OPR
Dog1	1350	SV, IPR, OPR
Dudek	1145	SV, OCC, DEF, FM,IPR, OPR, OV, BC
FleetFace	707	SV, DEF, MB,FM, IPR, OPR
Freeman1	326	SV, IPR, OPR
Girl	500	SV, OCC, IPR, OPR
Human8	128	IV, SV, DEF
RedTeam	1918	SV, OCC, IPR, OPR, LR
Singer2	366	IV, DEF, IPR, OPR, BC
Skating 1	400	IV, SV, OCC, DEF, OPR, BC
Singer1	351	IV, SV, OCC, OPR
Surfer	376	SV, FM, IPR, OPR, LR
Twinnings	472	SV, OPR
Toy	271	SV, FM, IPR, OPR
Vase	271	SV, FM, IPR
Walking	569	SV, OCC, DEF
Trellis	120	IV, SV, IPR, OPR, BC
Walking2	500	SV, OCC, LR

Table 1. Frame numbers and sequence attributes of 26 tracking sequences

Trackingalgorithm	Characteristic	Scaleadaptation
CSK	gray	no
CN	gray, color names	no
KCF	HOG	no
DSST	gray, HOG	yes
SAMF	gray, HOG, color names	yes
KCFDPT	HOG, color names	yes
fDSST	gray, HOG	yes
FSAKCF	HOG	yes
FSAKCF1	HOG	yes

Table 2. Characteristics of 9 tracking algorithms

Xuedong He, Shengzong Zhou. Fast Scale Adaptive Kernel Correlation Filtering Algorithm for Target Tracking[J]. Laser & Optoelectronics Progress, 2018, 55(12): 121501

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information