Fusion of Visible and Infrared Images Based on IHS Transformation and Regional Variance Matching Degree

6th Annual 2018 International Conference on Geo-Spatial Knowledge and Intelligence

Zhou, Chengquan^1^2 ; Zhao, Jinling^1 ; Pan, Zhenggao^3 ; Hong, Qi^1 ; Huang, Linsheng^1

National Engineering Research Center for Analysis and Application of Agro-Ecological Big Data, Anhui University, Hefei

230601, China^1

Beijing Research Center for Information Technology in Agriculture, Beijing

100097, China^2

Information Engineering Institute, Suzhou University, Suzhou

234000, China^3

关键词: Color distortions;    High frequency HF;    IHS transformation;    Intensity hue saturations;    Quantitative indicators;    Standard deviation;    Thermal infrared images;    Weighted averages;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/234/1/012021/pdf
DOI  :  10.1088/1755-1315/234/1/012021

来源: IOP

【 摘 要 】

It is usually necessary to identify and extract specific characteristics by deriving an informative fused image from multiple images. An effective fusion algorithm was proposed by combining the intensity-hue-saturation (IHS) transformation and the regional variance matching degree (RVMD) in our study. Visible and thermal infrared images of wheat were used as the original data sources. After finishing the IHS transformation, a fusion rule was designed to produce the new component I. More specifically, the high frequency fusion rule was generated by the RVMD with a threshold of 0.5 and a 3 x 3 moving window, and the weighted average was used as the low frequency fusion rule. Experimental results show that the proposed algorithm can avoid producing color distortion in comparison with the IHS transformation, and additionally, it can also enhance the edge contrast and produce more obvious texture resolution. In addition, three quantitative indicators including entropy, standard deviation and average gradient were used to validate the proposed algorithm. The analysis results show that the values of three indicators are respectively 7.82, 63.93, 10.06, which are better than the results derived from the IHS transformation and regional variance fusion.