关于图像分割的 C-V 模型改进算法研究

摘要

Active contour in image segmentation (active contour)or "snake" (snake)model can formulate the energy functional image segmentation problem as a minimization of a closed curve.In view of the existing problems in the C-V model,such as the existence of no obvious edge,the lack of a clear texture characteristics,the difficulty for GAC model to achieve segmentation issues,the energy functional model to derive global information of the image and the signed distance function are obtained through the evolution model of the existing various curve comparison analysis.This model integrates classical geometric curve model and the advantages of C-V model,and by introducing a penalty term,there is no need to repeat the update curve initialization,and the time step choice is increased.The model is better than the traditional C-V model in the convergence speed and accuracy.%在图像分割中的活动轮廓(active contour)或“蛇”(snake)模型可以将图像分割问题归结为最小化一个封闭曲线的能量泛涵。针对 C-V 模型中存在的问题,如存在没有明显的边缘,也缺乏明显的纹理特征,GAC 模型将难以实现成功的分割等问题,通过对现有各种曲线演化模型比较分析,得出一个图像全局信息和符号距离函数的能量泛涵模型。该模型综合了经典几何曲线模型与C-V 模型的优势,同时通过引入惩罚项,无需重复更新曲线初始化,并加大时间步长的选择。该模型比传统的 C-V 模型在收敛速度和精确度上都有了很大的提高。