主动轮廓小波多尺度变形边缘提取方法

Deformable Boundary Extraction Scheme Based on Active Contour Model and Wavelet Multi-scale Analysis

主动轮廓在图像分割和边缘检测中得以广泛应用,但其也受到多种因素的影响。针对图像轮廓周围其它图像信息对轮廓边缘的干扰,以及进化曲线难以到达深度凹陷的问题,提出了主动轮廓小波多尺度变形轮廓边缘提取方法,利用小波变换的多尺度分析来缩减或淡化非轮廓边缘信息,解决由于噪声以及轮廓周围其它图像信息对进化曲线的错误引导;通过对正则化力场进行改善,在进化曲线的动态轮廓搜索点上构建一自适应外力来引导凹陷处轮廓的进化方向;最后,给出了该方法的具体算法步骤。这些措施的实施,较好地解决了所研究的问题,实验表明该方法是可行和有效的。

Though active contour models are limited by many factors, they have been widely used in image segment and edge detection applications. A deformable boundary extraction scheme was proposed, it was for solving problems that contour was easily affected by the image information around edges, and deep concave area was difficult to arrive for the evolution curves. Firstly, the information which did not belong to the edges was decreased and controlled using wavelet multi-scale analysis, thus guide errors caused by noise and image information near the edges shouM be significantly corrected; and then, with modified regularization force fields, an adaptive external force was built at the searching points of evolved curves, which coud guide evolved curves convergenee to the concave boundary; finally the detailed algorithm was given. The experiment shows that the scheme is feasible and effective.