Image Fusion Technology Research

Image Fusion Technology, as an important part of the multi-sensor information fusion—the fusion of visual information, is a very popular research area in recent 20years. Image Fusion mainly aims to improve the reliability of images by processing the redundant information and the definition of images by processing complementary information among several images .In recent years, image fusion has become into a new important and useful technology in the field of image analysis and computer vision. This paper is mainly to discuss pixel-level image fusion technology.After introducing the background of research, basic concepts, development, internal and external applications and researches of image fusion technology, the paper elaborates the fundamental principles and methods employed in pixel-level image fusion technology. Then the de-noising, registration and fusion of image, which are the three steps of image fusion, have been deeply studied. And give the experimental data and result analysis. The main work of this these include:1. Image de-noising research: images devices in the process of photographing, will inevitably be subjected to noise interference, de-noising will have an important impact on the quality of the image and visibility of the image fusion. Through the traditional method of de-noising research and analysis, taking into account median filter method is simple and universal. Median filter is selected as a pre-processing method in image fusion.2. Image registration: a method of image registration based on feature extraction and hausdorff distance is proposed. At first, do Gabor convolution with the image, and then measure the energy formula to extract features, affine transformation formula's transformation coefficients are calculated by the feature point and Hausdorff distance, greatly increasing the precision, taking into account the faster, easier to design a strategy to achieve the search.3. Image fusion: through studying and analyzing fusion algorithms in the space domain and transform domain, this paper proposed a new image fusion method based on wavelet transform. To the different characteristics of the low-frequency components and high-frequency components, different fusion methods are selected. Finally, the fused image is attained by inverse wavelet transform. Experimental results show that the algorithm takes into full consideration the wavelet transform and HVS characteristics and enhanced the space details of images and fusion results.