A Study Of Fusion Method For Infrared And Visible Image

In recent years, with the increasing development of multi-sensor information fusion technology,the idea of multi-sensors’ joint work has come true. At the same time, with the increasing demand for the video image processing technology, many new sensors have emerged and their performance have become better than ever. Consequently, to integrate all the advantages of various image sensors becomes an important research field.The data fusion theory becoming more and more perfect, no matter at the military market or civilian market, image fusion technology has been applied in some significant areas. Among the image sensors, infrared image sensors play an important role in the military field. So the researchers put an emphasis on the infrared and visible image fusion technology. In 90 s last century, some developed countries, having started the related research and devoted abundant resources, finally achieved many meaningful production. Especially American has applied this technology in abundant weapons. Compared with foreign institution, we started studying the image fusion later lack of key technology. Currently, referred by the related statistic, only researchers from Beijing Institute of Technology and Nanjing Institute of Technology committed themselves to the study of image fusion technology, moreover they have developed their own real-time fusion processing system successfully which remained at the stage of experiment. Just few institutes or companies apply this technology to their product development. In this paper, the gray-level fusion technology of infrared and visible image is well studied. The author’s major contributions are outlined as follows:1. The sound fusion algorithms are simulated and analyzed. With the researcher’s continual study, there have emerged many fusion algorithms which are classified as gray-level fusion and color-level fusion. Pixel weighted method、Laplace pyramid and wavelet transform method are included in gray-level fusion, while MIT、TNO、color transfer method are included in color-level fusion. Some gray-level fusion methods are mainly studied in the paper.2. The fusion method based on wavelet transform is studied. We start the study of fusion method with wavelet transform algorithm on the basis of image multi-resolutionanalysis theory, and the infrared and visible image sensors’ different imaging features are emphatically considered. Comparison of experiments results reveals the defects of fusion method based on wavelet transform. Finally, we present an adaptive fusion method resulting in better performance.3. The fusion method based on image segmentation is studied. According to the problem needed to be solved of the method mentioned above, we study the clustering analysis theory, then present a region-based fusion method aimed at fusing the infrared interested target and remaining the nature scene of visible image. It proves that the infrared image’s unreal texture information has little influence on the fusion result and fusion result is greatly suitable for users’ observation.Finally, the paper summarizes the whole dissertation and gives a outlook to the next research of image fusion technology.