Study On Multifocus Image Fusion Using PCNN

The depth of field of imaging devices,such as light optical microscope,camera with finite depth of field and so on,is often limit.It makes people very difficult to get an image in which all the relevant objects is focused.To solve this problem,multi-focus image fusion is used to fuse all the source images which are focused from different points.In this way,all the focused objects in source images can be extracted to the fused image.Firstly,this paper reviews the current situation and level of image fusion.Then,the main technology and evaluation criterion of multi-focus image fusion are introduced.Next,two multi-focus image fusion methods are proposed: multifocus image fusion method based on pulse coupled neural network(PCNN)and guided filter(GF),multi-focus image fusion method based on PCNN and random walk(RW).Finally,experiments are done to evaluate the two multi-focus proposed image fusion methods respectively.Compared with image enhancement,multi-focus image fusion method is required to faithful to original images and preserve as much original information as possible.But on the other hand,the aim of multi-focus proposed image fusion and image enhancement has something in common: they all expect to acquire a clearer image than original image.The above mentioned connection between image fusion and image enhancement is used by multi-focus image fusion method based on PCNN and GF.The main step of the proposed method is that: Firstly,the source images are preliminarily fused by GF.Then,the preliminary fused image is enhanced by PCNN.Finally,fit the enhanced image with the source images to obtain the final fused image.The main step of image fusion method based on PCNN and RW is that:Firstly,the source images are preliminarily fused by PCNN.Then,the pixels of the fused image extracted from different source images are treated as different regions.Finally,these regions are modified to get the final fused image by RW.