Fusion Technology Of Infrared And Visible Images

Image fusion is aimed at exploiting redundancy data from multi-image to enhance the image'reliability and exploiting complementary information from multi-image to improve the image'definition. It is the advantages of image fusion that can expand the system'work scope and increase its reliability and cost performance ratio. Hence, image fusion technology has played a quite important role in remote sensing, medicine, natural resources exploration, landform analysis etc. This paper mainly studies the fusion technology of infrared and visible images. Mostly, fusion rules under the multiscale analysis, color image fusion algorithms and the selection of image fusion quality evaluation indices is studied.Multiscale analysis technology is often adopted in current image fusion algorithms. It contrasts pyramid transforms, wavelet transforms and beyond wavelet transforms in decomposition coefficients structural characteristics, transform effects, processing time and quantificational evaluation of reconstruct- ion images. And it compares compared Laplace pyramid transform, Contrast pyramid transform and Gradient pyramid transform. Moreover, it analyzes four kinds of lifting wavelet scheme and discrete binary wavelet transform. Besides, it compares Curvelet transform, Contourlet transform and nonsubsampled Contourlet transform. Expermental results highlight the advantages of beyond wavelet transforms which can capture curve's singularity. In addition, on the basis of multiscale analytical image fusion scheme, it compares fusion rules based on both single pixel and region, and contrasts fused images and performance indices by Wavelet and Curvelet transforms with different fusion rules. The results indicate that it can make better effects by using region-based on fusion rules.According to Piella multi-resolution analysis image fusion framework and human visual contrast sensitivity function, exploiting the second generation Curvelet transform, it proposes several fusion rules and seeks the optimal fusion rules on gray images fusion. And the proposed rules can obtain better visual effects than the existing fusion rules and accords with human visual perception. Furthermore, it exploiting the advantages of nonsubsampled Contourlet transforms whose coefficients have very strong dependency on among different scales, directions and positions, and the advantages of lαβcolor space which is perceptually decorrelated, it proposes a color image fusion algorithm. The algorithm, than traditional IHS color space transform and other color image fusion methods, can make fused image retain more spectrum information, target striking, color distortion less and abundant scene information. What's more, through the literature survey and experimental analysis, it classifies evaluation indices, and sums up the method of evaluation indices selection for different sensors images and on different applications.