Research And Implementation Of Seamless Image Composition Using Possion Blending

Natural image composition refers to matting or extracting interested foreground object from a source image and compositing it into a target image,while making the composited image be as visually pleasant as possible.Image matting and compositing is one of the most active research topics in the fields of digital image processing and computational photography.Since image matting and compositing can change the local content of digital image,it has wide application potentials in the fields of film production,virtual reality(VR)and augumented reality(AR).From the point of engineering practice,this thesis researches on the Possion blending based natural image seamless composition,which includes Alpha image from image matting and improved Possion image blending.Specifically,the main works and contributions are summarized as follows.First,an Alpha image matting approach is proposed,which is based on simplified intelligent scissor.The original intelligent scissor is simplified by simplifing its cost function and using optimized searching strategy.The searching speed for optimal path is improved about 5-6 times almost without sacrificing segmentation accuracy,which fully meets the requirements of initial object contour extraction for image matting.After obtaining the robust Matte,the Alpha Matting approach is exploited,which turns the problem of resolving the Alpha value of unknow region as a problem of graph labelling.Furthermore,random walking algorithm is used to resolve the issue of graph labelling,so as to obtain high-quality Matte.The experimental results on several typical images show that the proposed approach has achieved acceptale results.Secondly,an improved Possion image blending approach is proposed for image composition.It is built on the basis of Possion image blending.Firstly,an intuitive explanation is made about Possion image editting.In essence,the proposed approach is a combination of Possion blending and Alpha blending.The basic idea behind the proposed approach is that it ensures dependency on the boundary pixels of both target and source images,rather than just those of the target.Thus,the problems of color bleeding and bleeding artifacts are greatly reduced.The experimental results on several groups of typical test images show that the proposed approach has achieved seamless image compsition results,which are desired as our anticipations due to novisually annoying artifacts such as color bleeding.