Study On The Multi-focus Images Fusion Based On Wavelet Transform

Because of the short depth of the imaging device when the same optical sensors image in one area, it is difficult or impossible to make all the objects focused in the same scene. This problem can be solved by multi-focus image fusion which has attracted more and more attention in recent years, and widely used in these areas such as machine vision, digital camera and object recognition and so on. At the same time with the improvement of wavelet theory, the wavelet's multi-scale, multi-resolution characteristics in time-frequency make it a powerful analysis tool in image processing fields.This paper focuses on the multi-focus image fusion based on wavelet transformation, making in-depth discussion and improvement on the existing algorithms from the following two aspects, the wavelet transformation and the rules of multi-focus image fusion. Since the wavelet transformation has good localization properties in time and frequency domain at the same time, any of the details of the analyzed object can be focused on by taking on the high-frequency precise time or space step size. At present, however, there are two issues in the processing of image fusion by wavelet transformation: the selection of best wavelet basis function and wavelet decomposition level. In this paper, a large number of experiments are made to determine the optimal wavelet basis function, while with sub-image area ratio in wavelet decomposition and low-frequency band root-mean-square error criteria to determine the optimal wavelet decomposition level. This paper also makes improvement on the rules of multi-focus image fusion based on wavelet transformation especially on the aspect of the high-frequency coefficients which are reflected better by the image of the more obvious features of the frequency domain: regional contrast. With the rule:"high-frequency wavelet coefficients with large regional contrast indicate the clear part of the pictures", the fusion image is reconstructed. The experimental results show that, whether it is based on root-mean-square error, the average gray-scale error, entropy, space frequency of objective evaluation criteria, or subjective visual evaluation, the proposed selection methods are feasible and effective.The following main parts are included.Firstly, this paper discussed the basic theory, present state and means of image fusion technology, classified multi-focus image fusion algorithm and evaluation means of image fusion consequences.Secondly, this paper discussed the wavelet theory and the wavelet decomposition and reconstruction of image, studied the effects on selection of wavelet basis function and wavelet decomposition level to multi-focus image fusion, at last the best choices are proposed.Thirdly, a new method: regional contras of wavelet transformation are proposed for multi-focus image fusion based on wavelet transformation and Matlab is employed to simulation algorithm;Finally, this paper summarized the research achievement and insufficient of this paper and pointed out development direction.