Research On Infrared And Visible Image Fusion Technology

With the rapid development of multi-sensor imaging technology, multi-source image fusion has been the subject of extensive research and attention. As a hot branch of the field of multi-source image fusion, Infrared and visible image fusion technology plays an important role in military、medical、monitoring、biology and other fields. Infrared image reflects the difference of thermal radiation in surface of the object with all-weather feature, which makes infrared technology work well especially at night and it has been widely used in heat target detection. However, the background information in the infrared image is often vague, so it is difficult to get a full description of the scene; On the contrary, the visible imaging technique can accurately reflect the general information of the scene, but can’t detect the hidden target and difficult to work in the low light conditions. Thus, the research on the fusion of infrared and visible images with complementary feature has a comparative significance. This paper focuses on the infrared and visible image fusion technology, summarizes the technical difficulties and hot issues and related solutions, and then proposes some new solutions.First, study the current popular multi-scale analysis methods for image fusion extensively and depth, compare the advantages and disadvantages of different multi-scale analysis tools, identify multi-scale analysis methods which suitable for image fusion. On these basis, focus on the high-frequency and low-frequency fusion rule algorithm after multi-scale decomposition. According to the different features of infrared and visible light images, combined with the non-local mean idea and Gaussian blur, put forward an improved low-frequency fusion rules. In the high-frequency fusion rules, combined the fact that high-frequency coefficients express texture edge significance of image, the fusion rules with maximu absolution select is performed. Experimental results show that the algorithm can effectively keep the target information of infrared image and texture background information of visible images and it effectively suppress the effects of noise as a result of basing on regional integration and the idea of non-local means.In addition, In view of the target information of infrared image, draw the image saliency, which is a hot research direction in image processing, using saliency algorithm to calculate the saliency map of infrared image, so the saliency value of the target and background in infrared image can be calculated,and then guide the low frequency coefficient fusion rules by the use of saliency map. Image saliency quickly and accurately identify significantly goals effectively of image mainly through simulation of the human visual observation, and then provides more reliable image information for the subsequent image processing. Compared to traditional infrared and visible image fusion method and other segmentation, this approach can accurately highlight the target information of infrared image. At the same time, the influence of infrared background for fusion image has been effectively sharpness, subjective and objective evaluations of this algorithm are better.