Structure Continuous Seam Carving Algorithm For Image Resizing

With the rapid development of multimedia technology and video imaging technology, there appear diverse display devices, such as mobile phone, TV, Pad, laptop. Each Image has a fixed resolution, and when an image displayed on the display device which has different aspect ratio would cause a decline in image information and visual effect. Therefore, we need to use image resizing techniques to change the resolution of the image to adapt to the current display device, so that user can obtain accurately image information.The existing image resizing methods mainly exist in the following three problems:(1) Traditional image resizing methods, such as cropping and scaling, they didn’t judge the importance of the image information and would easy to cause the result of losing important image information;(2) For different input image, user need setting different parameters, which is too complicated to operate;(3) Seam carving sometimes break the structure continuous and cause distortion of the image. According to the above problems, we make a deep study on image resizing, and put forward the structure continuous seam carving algorithm for image resizing technical framework. The main content of the dissertation is as follows:(1) Put forward a new image size adjustment technical framework, namely the structure continuous seam carving algorithm technical framework. First use seam carving to find the optimized carve of input image, and separate the image into two pictures, we assume that picture lis smaller than picture2. Then use coarse to fine seam alignment method to align the seam and picture1. According to the alignment results, the small part have a deformation, splicing two pictures would obtain the image size adjustment results. The experimental results show that the structure continuous seam carving algorithm technical framework can solve the seam carving distortion of structure fracture and deformation phenomenon.(2) Put forward a coarse to fine seam alignment method. Method first used jointly by gradient, eight-neighborhood, RGB color constraint energy function characterized by significant points to find matching pixel level, for the whole correct matching build the foundation. and eight pixels according to color information segment matching pixel level, according to the characteristics of RGB color, coordinate information was not significant level find sub-pixel matching points. The experimental results show that the cutting seam alignment method to get high precision and accurate matching results.(3) Coordinate domain based Image deformation. Method first solve Poisson equation get the result of the cutting seam alignment offset the whole deformation area, and then through the principle of inverse mapping coordinates domain interpolation deformation results are obtained. The experimental results show that the image deformation method results without hole fracture phenomenon.