A multi-focus image amalgamation method based on imaging mechanism and nonsampled Contourlet transformation

Abstract

The invention discloses an image fusion method based on an imaging mechanism and non-sampled Contourlet transform. The method comprises the following steps of: first, carrying out the multi-scale and multi-direction decomposition to a source image and obtaining sub-band coefficients of different frequency domains via the non-sampled Contourlet transform; next, building a fusion rule based on the direction vector norm to the low frequency sub-band coefficients and building a fusion rule based on the union of the local directional contrast and the direction vector standard deviation to the band-pass direction sub-band coefficients; then, respectively combining the low frequency sub-band coefficients of the source image and each band-pass direction sub-band coefficients according to the built fusion rules and obtaining the non-sampled Contourlet transform coefficients of the fusion images; finally, rebuilding the fusion images via the non-sampled Contourlet inverse transform. The invention has the advantages of good fusion effect, little infection of the registration error on the fusion capability and being capable of effectively avoiding transferring the noise to the fusion images, and is applicable to the subsequent treatments and the image displays of the imaging systems.

Description

technical field [0001] The invention relates to the field of image processing, in particular to an image fusion method, which can be used to fuse multi-focus images to obtain an image with good global focus. Background technique [0002] Due to the limited depth of field of the optical lens, it is difficult for people to obtain an image in which all the scenes are in focus when taking pictures. One of the effective ways to solve this problem is to take multiple images of the same scene with different focus points, and then fuse them into one image in which all the objects in the scene are in focus. Due to different focus points, multi-focus images have different clear areas and blurred areas. The purpose of multi-focus image fusion is to select clear areas in each image and combine them into one image, while avoiding the introduction of false information. Multi-focus image fusion technology can effectively improve the utilization rate of image information and the reliabilit...

AI Technical Summary

Problems solved by technology

However, there are two main problems when wavelet is used for multi-scale analysis of images: ①Wavelet transform can only reflect the zero-wich heterogeneity of the signal, that is, it can reflect the position and characteristics of singular "points"

Therefore, when using two-dimensional wavelet to analyze the image, it is shown that "point" is used to approximate "line", which requires more coefficients to represent the straight line or curve in the image, so that the sparse representation of the image cannot be realized

② When analyzing the image signal using the two-dimensional wavelet generated by the one-dimensional wavelet tensor, only the information in the three directions of horizontal, vertical and diagonal can be obtained, and the direction of the edge in the image cannot be accurately expressed

Therefore, the image fusion method based on wavelet transform is easy to introduce "artificial" effect or high-frequency noise, which affects the quality of the fusion image to a certain extent.

[0006] (2) From the perspective of fusion rules: On the one hand, when most methods select low-frequency sub-band coefficients, they use an average method, which will reduce the contrast of the fusion image to a certain extent; for high-frequency sub-band coefficients The noise and the geometric features of the image are not distinguished during the selection, which makes it easy for the fusion method to extract noise as effective information and inject it into the fusion image

On the other hand, when formulating fusion rules, most fusion methods do not analyze the imaging mechanism of the source image, but only start with the image itself, and treat the image to be fused as an ordinary two-dimensional signal.

Therefore, the resulting fusion method does not achieve the best fusion effect

Images