In recent years,it has been difficult for imaging sensor with single-mode common lens to meet market demand,due to its small field of view,and lack of information to capture scenes.Multimodal ultra wide-angle imaging sensors,with wide field of view,can synthesize complementary information of various types of images,so as to make multimodal fusion images with richer scene information.Therefore,Multimodal ultra wide-angle imaging sensors have attracted more and more attention.Currently,due to their complementary imaging characteristics,infrared and low-light bands are widely used in multimodal image alignment and fusion.On the other hand,ultra wide-angle images usually suffer image aberrations,which cannot be applied directly and need to be corrected for image aberrations.In this thesis,the calibration,alignment and color fusion are studied for ultra wide-angle infrared images and ultra wide-angle low-light images.The detail contents are as follows:(1)Ultra wide-angle image distortion correction.Against ultra wide-angle image distortion,this thesis analyzes the classification of lens distortion and the principle of lens imaging.The grid template method and the equivalent spherical method are used for aberration correction respectively.Considering the defects of the above two methods,an improved hemispherical model algorithm is proposed,which reduces the correction error and improves the correction effect.(2)Alignment of infrared and low-light image.For the alignment of infrared and low-light images,this thesis firstly performs rough matching of image features based on the Grid-based Motion Statistics method.Then,a fine matching algorithm based on the combination of distance constraint and slope consistency is proposed to perform the initially screening of coarse matching feature point pairs for fine matching.Finally,the screened coarse matches are further selected by Random Sample Consensus algorithm to obtain the final feature matches.The image alignment algorithm in this thesis improves the image alignment accuracy.(3)Color fusion of infrared and low-light image.Aiming at the color fusion of infrared and low-light images,this thesis first uses multi-scale transformation to achieve gray-scale image fusion,which is used as the brightness component of the color fusion image.The pseudo-color fusion of infrared and low-light images is performed to enhance the fusion image levels.Then a natural color image fusion algorithm based on homomorphic filtering is proposed.The color components of pseudo-color images are improved by homomorphic filtering.The resulting histogram is matched with the color components of the reference image to obtain the color components of the color fusion image.Finally,color components together with the brightness component constitute a natural color fusion image.The color fusion algorithm improves the naturalness of color fused images without losing texture detail information. |