Open AccessImage Registration of Infrared and Visible Image Based on Geometric Constraint
Author(s) -
Juan Wang,
Kuo Hsing Cheng,
Cong Ke,
Min Liu,
Linkang Cai,
Hao Shi
Publication year - 2019
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1419/1/012008
Subject(s) - ransac , feature (linguistics) , artificial intelligence , computer vision , image registration , matching (statistics) , computer science , point (geometry) , constraint (computer aided design) , feature detection (computer vision) , pattern recognition (psychology) , point set registration , pixel , image (mathematics) , mathematics , image processing , geometry , philosophy , linguistics , statistics
With the wide application of infrared sensors and visible sensors, the registration of two heterogeneous images has caused widespread concern. Multi-angle comprehensive analysis of multi-sensor imaging systems can obtain more accurate image information. However, the main challenge of image registration is the detection of feature points and the acquisition of the correct feature point matching pairs. A Geometric constraint method is proposed to obtain the correct feature matching pairs. Three feature point matching pairs are randomly selected in the two images, and the ratio of the corresponding line segment lengths is obtained according to the coordinates of the three feature points to find the most three similar feature point matching pairs. Compared with the RANSAC algorithm, the method can obtain the correct feature point matching pair better. The algorithm runs on the DSP platform for about 30s, the error is 1 pixel, and the correct rate reaches 80%.