Study Of Image Fusion Technology Algorithm Based On Multi-scale Transform

With the development of the theoretical knowledge of image fusion and computer technology, image fusion technology has become more mature. Although people have had a wide range of research on multi-source gray image fusion, and achieved a more mature research results, yet the multi-focus color image fusion is studied relatively less.In this paper,we maily study the ways of multi-focus color image fusion and TM map and SPOT map fusion by using the multi-scale analysis based on wavelet decomposition.In this paper, we have conducted a lot of research for the multi-focus color images in RGB, HSV, HSIand YIQ three color space by using a two-dimensional discrete wavelet analysis, Two-dimensional discrete stationary wavelet, lifting wavelet. In the experiment, the color image is divided into three different color components in each color space, and then we can obtain respectively decomposed wavelet coefficients based on wavelet decomposition of each color component, and then use the same fusion rules, and finally get the fused image by the inverse wavelet transform and inverse color space transform. The study found that whether all of three color component or any each of color component is decomposed,fusion effect basically unchanged, so we can decompose any each of color component during the fusion to improve the efficiency of integration, and found that the multi-focus color image fusion based on the YIQ color space is better than on RGB or YIQ space. We also choose different fusion rules and different wavelet based on the YIQ color space to make researches, the experiment demostrates lifting wavelet is better for multi-focus color images than others.By comparison,YIQ space plus lifting wavelet plus the fusion rule1put forward in the paper can achieve a better fusion result.We also research SPOT map,a true color image and TM map,a grayscale image in RGB, HSV, HIS and YIQ color space. The TM map is divided into three different color components, and then the three color components of the wavelet decomposition and the wavelet decomposition of SPOT map were fused respectively to get three fusion coefficients, and finally get the fused image by the inverse wavelet transform and inverse color space transform. the study shows there will be a color distortion in the HSV,HSI and YIQ color space, while the fusion results in the RGB color space are poor.For this phenomenon, we designs a new fusion process for TM map and SPOT map fusion based on a two-dimensional discrete wavelet analysis in the HSI space.we first obtain the wavelet coefficients of the luminance component I of TM map, and is fused with the wavelet coefficients of TM map to get fusion coefficients,and use the fusion rule2put forward in the paper to get a new luminance component I, and finally get the fused image by the inverse wavelet transform and inverse color space transform.The study shows that the method is better in the effect of color fidelity, and better meet the visual requirements and subsequent computer processing.