Research On Fusion Algorithm Of Multi-focus Image

Multi-focus image fusion is to fuse the focus of different targets of the respective clear areas under the same scene to get a fully clear image.Due to the characteristics of multi-focus images,the fusion process reduces the introduction of irrelevant information,thus increasing the utilization of image information.Multi-focus image fusion technology can be widely used in computer vision,medicine,military and other fields.Through the in-depth study on the existing problems of multi-focus image fusion technology,we propose two algorithms.The main work of this paper is as follows:1.Due to the discontinuity at the boundary of the focus area,the pixels are erroneously selected in fusion processs,thus reducing the quality of the fused image.In this paper,a multi-focus image fusion algorithm is proposed based on NSCT and focus region detection.Firstly,the NSCT processing is performed on the image to get initial fusion image.The low frequency subband selection is taken by the method.The high frequency subband selection is taken by pulse coupled neural network(PCNN)method.Then,the decision matrix is obtained by morphological processing of the initial focus region.Finally,The ultimate fusion image is obtained according to the decision matrix.Experiments show that the proposed method provides high-quality fusion image and keep more detailed information of the image.2.In view of inadequate consideration of the spatial consistency and insufficient smoothness at the edge during the image fusion process,this paper presents a multi-focus image fusion method based on curvature filtering.Firstly,the source image is divided into the detail layer and the base layer by the curvature filtering method.Secondly,the standard deviation is taken and the consistency checked is performed at the detail layer.Thirdly,the base layer image is shared by the similarity region method.Finally,the fusion image is obtained by superimposing the fused detail image and the fused base layer image.The experimental results show that the edge smoothness of the image is well preserved by the curvature filter.