Image Registration Algorithm Based on Improved GMS and Weighted Projection Transformation

Fangjie Chen; Jun Han; Zuwu Wang; Guoqiang Zhang; Jianlian Cheng

doi:10.3788/LOP55.111006

Laser & Optoelectronics Progress, Volume. 55, Issue 11, 111006(2018)

Image Registration Algorithm Based on Improved GMS and Weighted Projection Transformation

Fangjie Chen^*, Jun Han, Zuwu Wang, Guoqiang Zhang, and Jianlian Cheng

Author Affiliations

School of Communication and Information Engineering, Shanghai University, Shanghai 200444, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(9)

Fig. 1. Distribution characteristics of correct matching and error matching

Download full size

Fig. 2. Correct matching and error matching feature for neighborhood score distributions

Download full size

Fig. 3. Nine lattice grids (purple) and five lattice grids (yellow) in a neighborhood sequence

Download full size

Fig. 4. Schematics of five lattice grid rotation. (a) Five lattice grids of the grid neighborhood; (b) rotate clockwise once; (c) rotate clockwise twice; (d) rotate clockwise thrice

Download full size

Fig. 5. Four groups of test images. (a) APAP trail images; (b) DHW temple images; (c) SPHP garden images; (d) proposed algorithm fan-sculpture images

Download full size

Fig. 6. Four groups of image stitching for comparison. (a) APAP trail image stitching; (b) DHW temple image stitching; (c) SPHP garden image stitching; (d) proposed stitching algorithm

Download full size

Table 1. Refined match pairs and running time

View table

Table 1. Refined match pairs and running time

Images	Algorithm	Rough match pairs	Refined match pairs	Time /s
Fig. 6(a)	GMS	10000	2685	0.065
Fig. 6(a)	Proposed	10000	2705	0.039
Fig. 6(b)	GMS	9889	3338	0.058
Fig. 6(b)	Proposed	9889	3420	0.034
Fig. 6(c)	GMS	10000	850	0.071
Fig. 6(c)	Proposed	10000	897	0.042
Fig. 6(d)	GMS	10000	23.58	0.045
Fig. 6(d)	Proposed	10000	2776	0.029

Table 2. Matching correct rate

View table

Table 2. Matching correct rate

Algorithm	Images
Algorithm	Fig. 6(a)	Fig. 6(b)	Fig. 6(c)	Fig. 6(d)
SIFT+RANSAC	97.32	98.74	97.65	96.51
SURF+RANSAC	95.62	96.68	96.13	95.71
BRISK+RANSAC	93.54	95.13	94.81	93.90
ORB+RANSAC	94.78	95.47	93.18	94.87
AKAZE+RANSAC	95.13	97.69	95.89	95.23
Proposed	97.57	98.61	98.14	96.56

Table 3. Root mean square errorpixel

View table

Table 3. Root mean square errorpixel

Algorithm	Images
Algorithm	Fig. 6(a)	Fig. 6(b)	Fig. 6(c)	Fig. 6(d)
SIFT+RANSAC	0.5189	0.3953	0.4571	0.4134
SURF+RANSAC	0.5854	0.4412	0.4925	0.4335
BRISK+RANSAC	0.7025	0.5633	0.6232	0.6069
ORB+RANSAC	0.6687	0.4951	0.5834	0.5488
AKAZE+RANSAC	0.6065	0.4238	0.5218	0.5047
Proposed	0.4527	0.3557	0.4079	0.3713

Tools

Get Citation

Copy Citation Text

Fangjie Chen, Jun Han, Zuwu Wang, Guoqiang Zhang, Jianlian Cheng. Image Registration Algorithm Based on Improved GMS and Weighted Projection Transformation[J]. Laser & Optoelectronics Progress, 2018, 55(11): 111006

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Image Processing

Received: Apr. 24, 2018

Accepted: May. 29, 2018

Published Online: Aug. 14, 2019

The Author Email: Chen Fangjie (amusi1994@shu.edu.cn)

DOI:10.3788/LOP55.111006

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology