Research And Application Based On Snake Model And Poisson Equation Mapping Algorithm

The foreground extractions of natural images are usually regarded as natural image matting by scholars.The effective matting technology can bring people very sense of interest from the original image is extracted, again through the technology of image synthesis, the resulting prospects figure is lifelike, seamless integration into another image in the background image to get specific effects.In our daily life, the technology of natural image matting is widely used, Such asapplying in the film special effects, the wedding process, the television media, advertising and so on.The way to achieve natural image matting is randomly selected a natural image, and manuallydivided the interesting part(called prospects graph), unnecessary area(called backgroundgraph) and the part between prospects graph and backgroundgraph(called unknown region) in picture, then separate the prospects graph form the whole imageto achieve their goals of the separation by the corresponding matting algorithm. The common matting algorithm including in Poisson matting? Bayesian matting,? Principal component analysis mattingand so on. It needs much manual intervention in the implementation process, the mistake of manual operation and inconspicuous backgroundwill lead to a high computational complexity of time. Active contour model (also called Snake) is an effective method using in image segmentation in recent years. Itonly simply setting the line of contour when we select the goal's graph, and automatically converge to the edge of target image according to internal and external's force by itself. It can effectively reduce human interaction and the computational complexity of time.The main work of this paper is the application of Snake model to the Poisson matting, a study by combining the Snake model with Poisson equation matting algorithm. First, A simple set of initial Snake contour lines in the foreground object, The characteristics of automatic convergence using Snake model, Through the elastic deformation of the template, Match with the local feature of the target image to achieve harmony, Make the contour line step by step closer to the true target profile; Secondly, The image in the contour line is used as the foreground, The image of the external contour line as the background, The prospect of the ratio of original image as the initial opacity value through the Poisson equation, Then Iterative calculation using Poisson equation, Final, it obtained before and background separation images, The use of image synthesis technology, The foreground image is fused to the selected background image. Last, the validity of the algorithm is verified by experiments.