带正则化拟合项的Chan-Vese模型

Chan-Vese Model with Regularized Fitting Term

为了提高Chan-Vese(CV)模型对噪声的鲁棒性,提出了一个带正则化拟合项的CV模型.新模型对CV泛函中图像像素灰度与拟合常数的距离函数进行磨光,在极小化新能量泛函的过程中,使得每个像素点的分割在其邻域范围内保持连续性,这样就可以排除噪声点的影响,获得正确的分割结果.并且采用一个基于新的边缘停止函数的测地弧长项作为正则项,进一步提高了模型的抗噪性.实验结果表明,对于不同噪声污染的人工图像和自然图像,本文模型都能取得较为满意的分割结果.并且对于强高斯噪声和椒盐噪声污染的图像,本文模型相对于经典的CV模型和LBF模型具有较大的优势.

A Chan-Vese （ CV ） model with regularized fitting term is proposed in this paper to improve the robustness of CV model to noise. The new model smoothes the distance between image pixel values and fitting constants in CV energy functional, and alms at keeping the continuity in segmenting each pixel within its neighboring block in the process of minimizing the new energy. In this case, it can overcome the influence of noise and obtain correct segmentation. In additional, a regularization term which is actually a measure of geodesic arc length based on a new edge stopping function is proposed to further improve the robustness of our model to noise. The experimental results on both noisy synthetic and real images show that the proposed model is robust to noise and can get satisfactory segmentation. In additional, the proposed model, compared with classical CV model and LBF model, can achieve better segmentation for the images corrupted by strong Gaussian noise or Salt and Pepper noise.