Image Fusion Method of Infrared and Optical Images Based on Target Extraction

2018 International Conference on Advanced Electronic Materials, Computers and Materials Engineering

材料科学;无线电电子学;计算机科学

Wang, Yu-Mei^1 ; Zhang, Ming-Yi^1 ; Li, Xiao-Ming^1 ; Wang, Tao^1 ; Huang, Ke-Ming^1 ; Wang, Dong^1

New Star Research Institute of Applied Technology, Hefei

230031, China^1

关键词: Canny edge detection;    High frequency HF;    Image fusion methods;    Laplacian Pyramid;    Mutual informations;    Region information;    Target extraction;    Threshold segmentation;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/439/3/032011/pdf
DOI  :  10.1088/1757-899X/439/3/032011

来源: IOP

【 摘 要 】

The fusion of infrared and optical images is difficult because images of the intended infrared targets are weakened by the background of optical images when guided by traditional pixel-based fusing rules. In this study, an improved fusion method based on infrared target extraction is proposed. First, to highlight infrared targets in the fused image, this method obtains the target's contour binary image by Canny edge detection and obtains the segmented binary image by using Otsu threshold segmentation method. The proposed method then fuses the two binary images using their complementary information to generate a more complete infrared target image which guides the fusion of low-frequency coefficients of the source images. Second, the extracted infrared target image, the source infrared image, and the V component of the source optical image are all decomposed by Laplacian pyramid transform. High-frequency coefficient fusion is performed using the rules of region mutual information, region match measure, and region energy, meanwhile low-frequency coefficient fusion is performed using the infrared target image and region information rules. Experimental results indicate that the proposed fusion method outperforms traditional fusion methods in preserving infrared target information in the fused image.