Complementary Tracking Algorithm with High-Confidence Updating Strategy Under Complex Scenes

Yaxiong Gu; Xin Li; Miaomiao Chen

doi:10.3788/LOP202158.1230004

Journals >Laser & Optoelectronics Progress >Volume 58 >Issue 12 >Page 1230004 > Article

Laser & Optoelectronics Progress
Vol. 58, Issue 12, 1230004 (2021)

Complementary Tracking Algorithm with High-Confidence Updating Strategy Under Complex Scenes

Yaxiong Gu¹, Xin Li^1、*, and Miaomiao Chen²

Author Affiliations

¹School of Mechanical and Electrical Engineering, Southwest Petroleum University, Chengdu, Sichuan 610500, China

²Hope College of Southwest Jiaotong University, Chengdu, Sichuan 610400, China

show less

DOI: 10.3788/LOP202158.1230004 Cite this Article Set citation alerts

Yaxiong Gu, Xin Li, Miaomiao Chen. Complementary Tracking Algorithm with High-Confidence Updating Strategy Under Complex Scenes[J]. Laser & Optoelectronics Progress, 2021, 58(12): 1230004 Copy Citation Text

show less

Fig. 1. Flow chart of our algorithm

Download full size

Fig. 2. Test results of different algorithms on the OTB2013 data set. (a) Precision; (b) success rate

Download full size

Fig. 3. Qualitative analysis results of 3 algorithms. (a) Boy; (b) Coke; (c) Human4

Download full size

Fig. 4. Qualitative analysis results of 6 algorithms. (a) BlurOw1; (b) Box; (c) FleetFace; (d) Liquor

Download full size

Specific configuration	Configuration data
CPU	Intel Core i7-7700HQ
Frequency /GHz	2.80
RAM /G	8
Operating system	64bit Windows10
Development tool	MatlabR2017a

Table 1. Experimental environment

Sequence	Frame	Attribute
BoyCokeHuman4BlurOwl	602291667631	SV, MB, FM, IPR, OPRIV, OCC, FM, IPR, OPR, BCIV, SV, OCC, DEF, MB, FM, OPRSV, MB, FM, IPR
Box	1161	IV, SV, OCC, MB, IPR, OPR, OV, BC, LR
FleetFace	707	SV, DEF, MB, FM, IPR, OPR
Liquor	1741	IV, SV, OCC, MB, FM, OPR, OV, BC

Table 2. Properties of the experimental video sequence

Algorithm	IV	OCC	IPR	SV	FM	MB	OPR	OV	DEF	BC	LR
SAMF	0.605	0.638	0.592	0.608	0.555	0.556	0.529	0.571	0.570	0.587	0.705
SRDCF	0.707	0.675	0.680	0.684	0.726	0.707	0.584	0.597	0.639	0.723	0.633
DAT	0.245	0.396	0.395	0.391	0.292	0.238	0.375	0.423	0.381	0.205	0.456
DATHOG	0.742	0.693	0.701	0.720	0.634	0.674	0.697	0.552	0.707	0.795	0.732
KCF	0.396	0.363	0.403	0.343	0.301	0.310	0.337	0.252	0.329	0.459	0.384
DSST	0.654	0.588	0.638	0.569	0.478	0.491	0.523	0.430	0.521	0.660	0.577
Staple	0.656	0.651	0.623	0.612	0.631	0.621	0.564	0.688	0.656	0.624	0.586
Ours	0.821	0.826	0.817	0.832	0.844	0.841	0.792	0.872	0.798	0.873	0.903

Table 3. Tracking precisions of 8 algorithms on 11 attribute sequences

Algorithm	IV	OCC	IPR	SV	FM	MB	OPR	OV	DEF	BC	LR
SAMF	0.526	0.557	0.533	0.510	0.467	0.470	0.449	0.492	0.491	0.533	0.559
SRDCF	0.617	0.587	0.612	0.612	0.676	0.632	0.503	0.528	0.549	0.624	0.604
DAT	0.222	0.307	0.291	0.292	0.274	0.251	0.286	0.353	0.324	0.197	0.230
DATHOG	0.632	0.607	0.591	0.653	0.691	0.620	0.571	0.552	0.582	0.672	0.586
KCF	0.323	0.271	0.336	0.261	0.282	0.284	0.261	0.241	0.270	0.401	0.258
DSST	0.569	0.492	0.539	0.473	0.426	0.427	0.423	0.391	0.469	0.589	0.439
Staple	0.588	0.570	0.545	0.524	0.553	0.529	0.478	0.507	0.590	0.576	0.466
Ours	0.691	0.673	0.682	0.668	0.719	0.732	0.650	0.736	0.635	0.758	0.587

Table 4. Tracking success rates of 8 algorithms on 11 attribute sequences

Algorithm	X_CLE /pixel	X_FPS(frame·s^-1)
SAMF	50.04	16.84
SRDCF	53.14	5.62
DAT	90.87	60.14
DATHOG	49.12	87.23
KCF	117.76	304.99
DSST	61.08	41.28
Staple	45.85	6.07
Ours	11.59	10.74

Table 5. X_CLE and X_FPS of different algorithms

Yaxiong Gu, Xin Li, Miaomiao Chen. Complementary Tracking Algorithm with High-Confidence Updating Strategy Under Complex Scenes[J]. Laser & Optoelectronics Progress, 2021, 58(12): 1230004

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information