Research And Application Of Image Resizing Method Based On Saliency

Many electronic devices such as high definition digital television, flat panel computer, PDAs, mobile phones provide image display functionality. The display devices present individuation and diversification. Original images usually have higher resolutions and different aspect ratios. Small screens cannot provide enough resolution to make the image meet the visual demand. Different aspect ratios may introduce image distortion.To solve these problems image resizing become a research hotspot.Recently, much work has been done on image resizing. Seam carving technique can efficiently inhomogeneously resize images, but its discrete nature may result in noticeably jagged edges of image objects after resizing.Non-uniform image wrapping is a continuous solution which overcomes the discrete nature of seam carving. However, the methods in propagate distortion along the resizing direction and the algorithm in can not preserve the large scale feature. We present a novel image resizing method based on mesh deformation which ensures that the shapes of prominent objects in target image are similar to their original shapes,and preserves important local regions and better diffuses distortion into less important regions. To accomplish this, we first establish the relationship between the characteristics of the source image and image saliency. Then we extract the mesh of important objects in the corresponding region of the image, and a mesh is defined over an image. In the initial mesh the similarity invariance of local region is calculated based on triangle similarity. A feature preserving mesh deformation to the target size is computed by minimizing an objective function about similarity. When resizing the image, we construct a quadratic energy to measure the similarity error for each local region. The resizing result is obtained by minimizing the energy function sum with salience as weight. The method is efficient enough for it only solves a system of linear equations. With the development of image resizing , digital television images can better adapt to the requirements of all kinds of the display screen, and digital television pictures meet the human visual sense.