Research On Image Edge Detection Based On Improved Snake Model

Image edge detection, one of the most important image analysis technology, plays an increasingly great role in image segmentation, image classification and recognition, computer vision and any other fields.Usually, traditional edge detection method can not satisfy the requirements which isn't included in the image itself. The active contour model (Snake model) not only contains the information of the image itself, but also adds prior knowledge from exports.It can accurately locate the boundary of the image.In this paper, we do a deep research on the edge detection theory which is based on the parametric active contour model and the geometric active contour model.Firstly, a review of image edge detection is presented. It includes the purpose, significance and the development of image edge detection.And then, the Snake model including its basic principle, and algorithm is introduced. Secondly, we discuss how to design the classic Snake model energy function. To solve the issue of the concave regional, we further introduce the basic algorithms and numerical implementation of gradient vector flow (GVF), explain the definition and principle of generalized gradient vector flow (GGVF) and take in an improved continuous energy function to improve the uniformity of contour points and make the curve smoother than before. Thirdly, we focus on discussing the Snake model based on the level set method, which is a geometric active contour model.The level set method consists of the traditional level set method and the level set method without re-initialization.The biggest difference between the two models is that the latter completely eliminates the need of the costly re-initialization procedure and therefore speeds up the curve evolution.In this paper, we using the method of combining morphological reconstruction method with level set method to reduce the noise in the image and improve the computational efficiency. Finally, we summarize the advantages and disadvantages of the parametric active contour models and the geometric active contour model respectively.