Image Stitching Based on CLAHE and Improved ZNCC

Guanqun Huo; Jinbo Lu; Shengxiang Luo

doi:10.3788/LOP202259.1211003

Journals >Laser & Optoelectronics Progress >Volume 59 >Issue 12 >Page 1211003 > Article

Laser & Optoelectronics Progress
Vol. 59, Issue 12, 1211003 (2022)

Image Stitching Based on CLAHE and Improved ZNCC

Guanqun Huo, Jinbo Lu^*, and Shengxiang Luo

Author Affiliations

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

show less

DOI: 10.3788/LOP202259.1211003 Cite this Article Set citation alerts

Guanqun Huo, Jinbo Lu, Shengxiang Luo. Image Stitching Based on CLAHE and Improved ZNCC[J]. Laser & Optoelectronics Progress, 2022, 59(12): 1211003 Copy Citation Text

show less

Fig. 1. Histogram clipping

Download full size

Fig. 2. Interpolation operation

Download full size

Fig. 3. Process of enhancing image contrast

Download full size

Fig. 4. After contrast enhancement, feature point extraction and matching results. (a) Traditional method; (b) proposed method

Download full size

Fig. 5. Schematic of sampling window rotation

Download full size

Fig. 6. The first and last images of the tested image sequence. (a) bark group; (b) leuven group; (c) bikes group

Download full size

Fig. 7. Comparison of number of correct matching points in the bark group obtained by different algorithms

Download full size

Fig. 8. Comparison of number of correct matching points in the leuvne group obtained by different algorithms

Download full size

Fig. 9. Comparison of number of correct matching points in the bikes group obtained by different algorithms

Download full size

Fig. 10. Original images of mosaic image. (a) The first group; (b) the second group; (c) the third group

Download full size

Fig. 11. The first group. (a) Direct extraction; (b) contrast enhanced by the proposed algorithm and then extraction; (c) screening by the proposed algorithm

Download full size

Fig. 12. The second group. (a) Direct extraction; (b) contrast enhanced by the proposed algorithm and then extraction; (c) screening by the proposed algorithm

Download full size

Fig. 13. The third group. (a) Direct extraction; (b) contrast enhanced by the proposed algorithm and then extraction; (c) screening by the proposed algorithm

Download full size

Fig. 14. Image mosaic results of the first group. (a) Classical SIFT algorithm; (b) algorithm in Ref. [2]; (c) proposed algorithm

Download full size

Fig. 15. Image mosaic results of the second group. (a) Classical SIFT algorithm; (b) algorithm in Ref. [2]; (c) proposed algorithm

Download full size

Fig. 16. Image mosaic results of the third group. (a) Classical SIFT algorithm; (b) algorithm in Ref. [2]; (c) proposed algorithm

Download full size

Window size	Threshold 0.8		Threshold 0.7		Threshold 0.6		Threshold 0.5		Threshold 0.4
Window size	$n_{c}$	$P_{c o r}$ /%	$n_{c}$	$P_{c o r}$ /%	$n_{c}$	$P_{c o r}$ /%	$n_{c}$	$P_{c o r}$ /%	$n_{c}$	$P_{c o r}$ /%
$3 \times 3$	210	95.89	286	96.95	339	97.13	388	97.24	420	97.45
$5 \times 5$	251	98.05	364	97.59	441	97.78	478	97.75	506	97.68
$7 \times 7$	280	97.22	408	98.08	481	98.16	530	97.79	546	97.67
$9 \times 9$	286	97.95	437	97.98	516	98.10	547	97.85	562	97.91

Table 1. Comparison of number of correct matching points and correct matching rate obtained by using different window sizes and different thresholds

Data set	SIFT	SURF	ORB	Ours
bark 1&2	99.62	92.68	99.06	99.76
bark 1&3	100.00	96.61	100.00	100.00
bark 1&4	100.00	89.58	99.75	100.00
bark 1&5	99.78	88.64	100.00	100.00
bark 1&6	100.00	60.00	100.00	100.00
leuven 1&2	98.77	94.66	99.06	99.40
leuven 1&3	98.36	88.66	99.00	99.39
leuven 1&4	97.96	90.79	99.10	99.13
leuven 1&5	97.45	81.11	99.06	98.21
leuven 1&6	95.02	79.21	99.09	97.78
bikes 1&2	98.25	96.14	99.06	98.32
bikes 1&3	96.43	94.96	98.27	98.36
bikes 1&4	97.02	90.61	98.17	100.00
bikes 1&5	94.27	86.99	95.77	95.00
bikes 1&6	88.66	84.00	90.55	90.25

Table 2. Correct matching rate of each group of images under different algorithms

Date set	Direct extraction	After enhancement	After screening
Group 1	67.86	68.77	70.74
Group 2	93.97	92.91	95
Group 3	82.89	91.87	97.29

Table 3. Correct matching rate of each group of images before and after using proposed algorithm

Guanqun Huo, Jinbo Lu, Shengxiang Luo. Image Stitching Based on CLAHE and Improved ZNCC[J]. Laser & Optoelectronics Progress, 2022, 59(12): 1211003

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information