Image Fusion Algorithm Based On The Non-aiiasing Contourlet Transform And Denoising Technology Research

Images, as a significant resource of intelligence, have been one of the most important source of cognition, back to ancient times. Image processing is widely used in satellite remote sensing, machine vision, biomedical, traffic, public security, military, and other fields, and already attracted broad attention, under half a century of development, Image analysis technology emerged in the 1960 s, after more than half a century’s development, achieved quite a number of achievements. As a two-dimensional information, the complexity of image decided the difficulty of image processing. The way of quickly, accurately and efficiently image information extracting, key information tracing, has been a hot research topic in the field of image. How to search for the key information in massive amounts of pictures and make an accurate rendering, is increasingly concerned.Image information processing has experienced the development from airspace to frequency domain, from Fourier transform to wavelet transform. These arithmetics have the same purpose, looking for the most sparse image feature representation, to capture the features of edges and textures in higher dimensional singular. Fourier transform and wavelet transform can not achieve the sparsest representation of image features, therefore, multi-scale geometric analysis was presented. Multi-scale geometric analysis offered a new way of thinking for image analysis, processing, and mining. Typical multi-scale transform includes Ridgelet, Curvelet, Contourlet, Bandelet, Directionlet, Shearlet. With anisotropic basis functions, they are able to decompose and approximate image information comprehensively. Relevant mathematical theory is still in development, there are still amount studies to research.In this paper, we design and realize two image fusion and one denoising algorithms based on non-subsampled Contourlet transform, weighted average algorithm and principal component snalysis. And then we make some primary discussion of denoising performance evaluation.We organize the paper as below:The first chapter introduces the concept, definition, application and performance of wavelet transform. The second chapter introduces the constitution, application and performance of Contourlet transform. The third chapter introduces the analyzation of frequency aliasing of Contourlet from the perspective of mathematics. The forth chapter realize two image fusion algorithms based on non-subsampled Contourlet transform, weighted average algorithm and principal component snalysis, and analyze the fusion results. The fifth chapter realize the threshold denoising algorithms based on non-subsampled Contourlet transform, and analyze results. The last chapter is the conclusions, summarization of our work and future study.