基于多尺度图像的主动轮廓线模型

ACTIVE CONTOUR MODEL BASED ON MULTI SCALE IMAGE ANALYSIS

主动轮廓线模型是广泛应用于数字图像处理的一种目标轮廓跟踪算法 ,但在实际使用过程中 ,现有模型易受干扰噪声及虚假边缘的影响 ,且对凹陷轮廓的跟踪能力较差 .在多尺度图像分析的基础上 ,引入梯度矢量流的概念 ,并改进其计算方法 ,提出了一种新的主动轮廓线模型 .该模型利用梯度矢量流产生的引力 ,在图像的尺度空间中搜索目标轮廓 ,不仅能有效地排除干扰 ,搜索凹陷轮廓 ,而且便于引入新的约束条件 .实验表明该模型有较好的鲁棒性和实用性 ,适用于噪声干扰情况下提取具有凹凸特征的目标轮廓 .

Active contour model is a widely used algorithm to track object contours in the field of digital image processing. But in the practice of application, the present models often subject to the influence of noises and fake edges, and their abilities of searching concave contour are not good. Based on the multi scale image analysis, the concept of gradient vector flow (GVF) is introduced and its algorithm is improved to present a new active contour model. The new model uses the forces produced by GVF to search object contours through the scale space of an image. It can efficiently get rid of the influence of noises and search concave contours. Furthermore, it is convenient to add new constraints to the model. Experiments prove that the new model is robust and practicable, and it is suited for searching concave or convex contours against contaminated background.