Multi-focus Image Fusion Based On Adaptive Region And Spiking Cortical Model

In the process of the actual object imaging,due to the limited focuse range of optical imaging system,so it is hard to present in the same image that focuse part and non-focuse part all clearly.Multi-focus image fusion technology can process multiple focus images that contains different clear objectives in one scene,and get a fusion image of all goals are clearly.This dissertation focuses on: multi-focus image is decomposed by spatial domain or transform domain algorithm,different fusion rules are used according to different information contained in the decomposed images,the fusion rules adopted in low frequency image and high frequency images are adaptative region and spiking cortical model(SCM)respectively.The main contents and innovations are as follows:1.In spatial domain,the cartoon-texture decomposition algorithm is used to decompose the source image.The cartoon-texture decomposition algorithm is used to divide the source image into cartoon and texture components.The cartoon component contains the main geometric structure,and it is the low frequency component of the source image;the texture component contains the main texture,fine detail,oscillation part and noise,and it is the high-frequency component of the source image.According to the main different information,regional variance is used in cartoon component to compare the adaptive region and in texture component to stimulate SCM.The simulation results demonstrate that the multi-focus image fusion algorithm using adaptive region and SCM based on cartoon-texture decomposition can preserve the details of the source image and the edge information well.2.In transform domain,NSCT and NSST transform algorithm are used to decompose the source image.In the process of image fusion,the information extracted from the source image is incompleted by NSCT decomposition,so it’s easy to generate various degree blurs when fused image in the position of details.Aimed at the lack of NSCT transform,this paper studies multi-focus image fusion using adaptive region and SCM based on NSCT domain,which can effectively extract useful feature information and get a good fusionimage in the fusion process.However,NSCT has high computational complexity,so NSCT does not apply to the real-time requirements of the occasion.NSST has a more flexible structure,so it can speed up the calculation.After the NSCT or NSST transform,the high frequency and low frequency coefficients are calculated by using the edge of energy,then using the edge of energy as the measurement method of the adaptive region and the stimulus input of the SCM respectively.The simulation demonstrate that,the multi-focus image fusion algorithm using adaptive region and SCM based on NSST domain can better preserve the details and the edge information of the source image.This method is superior to the current mainstream fusion method compared from the objective evaluation and subjective vision.