Research And Application Of Image Segmentation Algorithm Based On Snake Model

Image segmentation is the basic technique in image processing and computer vision fields, and also is the important part of most image analysis and computer vision system. With the increasement of oil and natural gas pipelines service time, the phenomenon of pipeline corrosion increases and wall thickness thins. If these defects aren't found and dealt with in time, it may lead to accidents and unnecessary losses. In short, the defects of pipelines may happen in manufacture operation and other aspects. Therefore, it is necessary to detect pipeline wall and extract the contour of injury region in time in order to prevent unnecessary losses.Active Contour Model, or Snake model, which was first introduced by Kass, has been used extensively in many applications of computer vision and image processing, such as edge detection, image segmentation and motion tracking, particularly to extract object boundaries. With the introduction of high-level information, Snake model has more satisfactory effects than what gained by the traditional method when dealing with disconnected edges in the image. This paper analyzes the principle and methods of parameter active contour models, including the minimum energy formula of active contour model, the physical significances, external design and numerical implementation. By experiments, the paper compares performance, advantages, disadvantages and the result of convergence when setting different initial outline among these models. The experiments show that GVF Snake model has bigger capture domain than other parameter active contour models such as traditional model and balloon snake model. In addition, it is insensitive to disconnected edges, can detect sunken part of image and has a high speed of convergence. But GVF Snake model has some defects, for example it is sensitive to weighting factor and has low speed of convergence in sunken part of image. Therefore, an improved GVF Snake model is achieved. Experimental results show that the improved GVF Snake model has a better force field and faster speed of iteration than GVF Snake model. We can obtain better segmentation results of equipment wall by GVF Snake and improved GVF Snake model. It provides a good environment for three-dimensional reconstruction and calculating the geometric parameters of the damaging parts of device wall image.