基于局部图像拟合偏差的活动轮廓分割模型

Active Contour Segmentation Model Based on Local Image Fitting Bias

针对灰度不均匀图像的分割问题,提出一个基于区域的活动轮廓模型。通过构造包含图像局部信息的局部图像拟合偏差能量泛函,度量真实图像与拟合图像的偏差,并在全局凸分割的基础上,将分裂Bregman技术应用到模型能量泛函的最小化问题中,以提高分割速率。同时引入边界检测函数更加准确地探测边界位置,以提高模型的分割准确性。实验结果表明,该模型不仅可以正确分割灰度不均匀图像和受噪声干扰的图像,而且对于多目标图像以及灰度分布均值相同、方差不同的图像,也能快速、准确地得到分割结果。

In order to segment images with intensity inhomogeneity, a new region-based active contour model is proposed. By introducing the Local Image Fitting Bias（L!FB） energy function that embeds the image local information, it can measure the difference between the original image and the fitted image. Moreover, based on the globally convex segmentation method, the split Bregman technique is applied to minimize the proposed energy function more efficiently. By using an edge detection function to the proposed model, the algorithm can detect the boundaries more accurately. Experimental results show that the proposed model not only can segment images with intensity inhomogeneity and images corrupted by noise, but also can efficiently and accurately segment multi-object images and images with similar intensity means but different variances.