A Improved GVF Snake Model Combined With Corner Detection

Digital image edge detection plays an important part in image analysis, such as image segmentation, interested region recognition and region shape extraction. As the base of digital processing, edge detection is the active topic in the machine vision research field. And there are many methods are applied in the edge detection. Such as differential algorithm base on spatial, Wavelet Transform, Mathematics morphology, methods based on energy. While Most of the traditional algorithms have some shortcomings, for example they are sensitive to noises; the detection results are not well. The Snake model proposed by Kass has great changes compared with other algorithms. It has a great visual effect, but also with its own disadvantages. It becomes very week where the edge is in a sunken area, especially the "U" and "Ω" shape. Resent years some improvements are take on the Snake model, and works well in a certain extent. The GVF Snake is a typical improve model, the result are well but also have some space to be improved in convergence to sunken edge.In this paper, we do a deep research on the edge detection theory and algorithm, base on analyzing the traditional algorithms in detail; we focus on GVF Snake model, and take the corner detection methods in. Base on the theory of Harris and SUSAN corner detection algorithms, this paper proposed a new corner extract method called Harris-USAN for the GVF Snake. The new corner algorithm combined with the GVF Snake is performs well as we implement in MATLAB. The experiment results show that the new GVF Snake model based on Conner detection are better than the original GVF Snake model.