Research Of Image Fusion Algorithm Based On Second Generation Wavelet Transform

Image fusion refers to the integration and extraction of two or more images, to obtain the same scene or target is more accurate, comprehensive and reliable images, making it more suitable for human eye perception or computer-processing. It is usually applied to de-noising, improve resolution, increase the amount of information, and improve clarity, to monitor changes in the scene or object and the use of other sensors to replace or compensate for the image sensor in the image of lost or fault information. Image fusion technology can be effectively used with different input channels of visual information, complementarily and redundancy, eliminating multi-source sensor data that may exist between the redundancy and contradictions, and enhance the transparency of information image, improve the signal to noise ratio and accuracy of the interpretation.The work area of image fusion is divided into spatial domain, spectral domain, frequency domain and the domain-scale integration. From low to high it also could be divided into pixel-level, feature level and decision level fusion at three levels,based on information of different degree of abstraction.Pixel-level integration is the most basic integration, feature-level fusion and the basis of decision level fusion, now has become the hot topic of research.As a kind of pixel level fusion methods, wavelet transform with its multi-resolution analysis and multi-scale analysis of properties, are currently being widely used in image fusion technology. Usually the structure and properties of the wavelet Fourier analysis are carried out under the framework, Swelden and others studied the use of lifting scheme wavelet constructed in the space domain issues, and has been based on lifting scheme of wavelet signal decomposition and reconstruction calculation method, this method is also known as second generation wavelet transform theory. Compared with the traditional wavelet transform which can only be carried out by frequency domain decomposition, the second generation wavelet transform is carried out in the time domain, therefore at present has become a new hotspot.This paper mainly studies the following aspects:(1) In-depth study of existing image fusion algorithm based on the traditional horizontal comparison of the wavelet transform and the second generation wavelet transform based image fusion algorithm and key steps in the content, pointing out that the second generation wavelet transform in image fusion algorithm is in the superiority of the image fusion.(2) Through studying a variety of wavelet bases to enhance programs and improving their performance after treatment compared with the wavelet, from which selected a more suitable for image decomposition and integration of the wavelet decomposition of the image.(3) For low-frequency and high-frequency coefficients after the decomposition of the image to reflect the characteristics of the different elements proposed low-frequency coefficients of the use of regional energy integration, high-frequency energy analysis combined with neural network clustering algorithm to improve the integration and use the Matlab simulation algorithm of the process simulation tools obtained the results of fusion images.(4) Through the effect of image quality evaluation, we can see that the algorithm in this paper obtained the image redundancy is small, fast computation time, small memory, high-resolution image content.