Research On The Multi-focus Images Fusion Based On Contourlet Transform

Image fusion in essence is to process a multiple source images contained complementary information and redundant information to obtain more accurate and a wealth of information, according to certain rules or algorithms. That can generate a synthetic image with the new space, spectrum and the time characteristics, so that fusion image are more conducive to people of their analysis and understanding. Multi-focus image fusion is based on the image fusion of different target-specific focusing on the integration of multiple images. Image fusion technology has been widely used in remote sensing image analysis, machine vision, medical diagnostics, target identification and other fields.At present, there are some problems about the wavelet transform used at pre-processing stage in the image fusion and the rule about selection coefficients of fused image, so it is difficult to accurately extract the source image of the clear area. In this paper, multi-focus image fusion algorithm based on contourlet transform has been studied. The paper by analyzing four aspects of Contourlet transform theory, fusion rules, image fusion experimental results and the figures on the expression of fusion characteristics indicates that the algorithm in this paper is better than the algorithm based on wavelet transform, especially the root mean square error and entropy of the image is better than the comparative literature's method.This article mainly contains the following:First of all, this article introduces the basic theory of wavelet transform and discusses the defects of wavelet transform applicationing in image processing. Then it introduced the theory and the advantages of Contourlet transform, which is more suitable for image analysis. Compared with wavelet transform, Contourlet transform is a new multi-resolution, multi-dimensional image analysis method that can represent the high-dimensional functions containing singular surface well and has better edge features for the image. The wavelet transform can only be good for the one-dimensional signal processing and it shows a serious redundancy for the two-dimensional signal. Contourlet transform finally uses the base structure which is similar to the line segments to approximate the original image, that is why it is called Contourlet transform.Second, a variety of image fusion algorithms are presented. This paper proposed a new image fusion algorithm based on Contourlet transform, which uses improved spatial frequency as a parameter to select the source of image data. This fusion method can retain a good detail and edge features for the source image. And the optimal number of decomposition level and the direction about Contourlet transform are studied through Matlab simulation. This section is improtant for this paper.Finally, this article discusses several commonly used evaluation means for image performance. Through comparison of the new image fusion algorithm improved by this paper and others, it shows that the new image fusion algorithm not only achieves a good visual effect, but also improves the edge and texture features for the image. At the same time, it retains more of the source image information, and the evaluation parameters is also superior to other algorithms.