Research On Application Of Shearlet Transform In Image Edge Detection

Edge is one of the fundamental characters of an image. Edge detection is one of themost important studying projects in computer vision and image processing. It is a hotpoint for researehers in the field of image measurement technology. Currently, manyalgorithms have been proposed for edge detection of images. But the existing theoriesand algorithms for edge detection still have some drawbacks and cannot detect edgessatisfactorily in some cases. Wavelet transform has good multiscale analysis and “time-frequency” technology, so it provides an effective way for image edge detection.However, despite its remarkable success in dealing with pointwise singularities, wavelettransform fails to provide an effective representation for images that contain curvesingularities. Shearlet transform is a new multiscale geometric analysis tool, it canovercome the shortcoming of wavelet transform which can not capture the geometryfeatures of multidimensional data, and two-dimensional images can be bestapproximation by shearlet transform. We propose two novel edge detecion algorithmsbased on shearlet transform and traditional edge detecion operators.In the first part of this paper, the research background of shearlet theory, itsdevelopment histories and its current research situation are overviewed. The conceptsand properties and the applications in image processing of shearlet are outlined.And then, canny criterion and canny edge detection algorithm are discussed, cannyedge detection algorithm realization process are described, the advantages anddisadvantages of traditional canny algorithm are analyzed, based on this, The new edgedetection method based on shearlet transform and improved canny operator isproposed, and the experiments indicate that the proposed edge detection method is akind of effective edge detection method. Finally, by the study of the imaging characteristics of medical CT image, a newedge detection method based on shearlet transform and LOG operator is proposed.This method is applied to medical CT image edge detection, and the MATLABsimulation results indicate that the method is effective.