Research And Implementation Of Bionic Algorithm For Color Fusion Of Infrared And Low Light Level Image

Nowadays,infrared and low light level image fusion is mostly in the stage of gray image fusion.Well,the ability of human eyes to distinguish gray information is much lower than color information.Color fusion images can make human eyes identify targets and scenes faster.Therefore,it is very important to carry out the research on color fusion algorithm of infrared and low light level image.Evolving billions of years,biological vision can accomplish many unimaginable visual tasks in machine vision and image processing.By studying the visual characteristics of limulus and rattlesnakes,this paper has proposed bionic algorithm to enhance and fuse the infrared and low light level image,thus obtaining color fusion images with natural color and outstanding target.The main research work of this paper are as follows:(1)Analyzing the feature of infrared and low light level image and introducing some commonly used color space model and its' interconversion for image color fusion,which lays a theoretical foundation for image enhancement and color fusion;Four classic infrared and low light level image color fusion algorithms are deeply analyzed and simulated,and their advantages and disadvantages are compared.(2)A bionic enhancement algorithm for infrared and low light level images is studied.In view of the low brightness level of infrared and low light level image,the logarithmic function which conforms to the visual characteristics of the human eyes is improved to adaptively adjust the brightness level of the image.Then,in view of the poor contrast of infrared and low light level image,the limulus eyes' lateral inhibition is improved,which can adaptively adjust the lateral inhibition coefficient according to the edge information of the image,as well enhance the local details and improve the image contrast.Finally,by comparing with the limulus eyes' lateral inhibition algorithm and Retinex enhancement algorithm based on histogram equalization,the results show that the bionic image enhancement algorithm can obtain better enhancement effects.(3)Because the Massachusetts Institute of Technology(MIT)bionic rattlesnake color fusion method(MIT method)only simulates the IR-depressed visual cell,and the target of the fusion image is not prominent,so an improved MIT algorithm is proposed in this paper,which can fully reflect the "enhanced" and "depressed" functions of the rattlesnake bimodal cells.Compared with MIT method,the color fusion image obtained by the bionic fusion algorithm has more prominent target and easy to be recognized by human eyes.(4)The subjective and objective quality evaluation are carried out by the proposed method,MIT method and Wang Yong method of Jilin University.The result of subjective evaluation shows that the color fusion image obtained by the proposed method is richer in color and has prominent target.The objective evaluation results show that the fusion image obtained by the proposed algorithm are higher than that of MIT algorithm and Wang Yong algorithm in terms of color degree,mutual information and structural similarity index measurement.It shows that the fusion image obtained by the improved MIT algorithm has more color information than MIT method and Wang Yong algorithm,it contains more source image information and is more similar to the source image structure as well.