Research On New Techniques For Digital Image Segmentation And Warping

Image segmentation and warping techniques are important in digital image processing, and are also fundament to object-based image editing system. In this paper, we firstly introduce the basic theory of fuzzy mathematics and fuzzy connectedness, then we present a modification to the fuzzy connectedness image segmentation. For the obtained region contour, we present a method to fair and fit it by some curve segments. Several image warping models are presented in the last. The main contributions are listed as follows:(1) Fuzzy Connectedness Based Image Segmentation and Its AlgorithmThrough checking the property affinity between the seed pixel and the pixel along the optimal path which has the largest fuzzy connectedness from the pixel to seed pixel, good results can be achieved especially for those objects with blurred boundary. Additionally, an image-scanning mechanism algorithm for detecting optimal paths is proposed, which calculate the fuzzy connectedness between pixels and the seed pixel one by one. This algorithm can make the best of the properties of fuzzy connectedness and property affinity and detect the optimal path between two pixels effectively.(2 ) A fairing algorithm to image contourUsing the concept of "point classifier", the points on the image contour are classified into three groups: faired points, corner and noised points, then we fair the noised points with bilateral algorithm. With the new method, features preserving and controlled shrinkage results can be obtained effectively.( 3 ) An effectively fitting algorithm to image contour by curve segmentsBased on the dominant points detected by some dominant detectors, a local optimal fitting result can be obtained by curve segments using vertex adjustment concept and dynamic programming algorithm. An automatic vertex inserting algorithm is presented to meet the user's need.(4) Some models to image warpingWe present some models to image warping: line-based locally constrained single point image warping, line-based locally constrained multiple points image warpin g and curvilinear-based locally constrained single point image warping. In the same time, a condition to control overlap in line-based locally constrained single point image warping is presented. These methods permit the users to interactively adjust the deformation with direct manipulation. Experimental...