结合全局和局部信息的水平集图像分割方法

发布时间：2018-04-26 13:39

本文选题：图像分割 + 图像噪声　；参考：《计算机应用研究》2017年12期

【摘要】：LBF模型对初始轮廓大小和位置非常敏感,并且只考虑了图像的局部信息,没有考虑图像的全局信息;CV模型利用图像全局信息,对初始轮廓具有较强的鲁棒性。两种模型对椒盐噪声污染的图像不能取得令人满意的结果。针对以上问题,在原有CV模型和LBF模型能量函数基础上,各自构造一个新的能量拟合项,增强对高斯噪声和椒盐噪声的抗噪性。采用新构造的CV模型,使用图像的全局信息得到粗分割轮廓;以粗分割轮廓作为新构造LBF模型的零水平集,利用图像的局部信息得到图像的精确分割结果。同时提出一种新的边缘检测算子,重新定义边缘停止函数,进一步提高模型的抗噪性。相较于CV和LBF模型,结合全局和局部信息的Wang和Qi模型,提出的模型能得到更优的图像分割结果,具有较强的抗噪性。
[Abstract]:The LBF model is very sensitive to the size and position of the initial contour and only considers the local information of the image. The CV model is robust to the initial contour by using the global information of the image without considering the global information of the image. The two models can not obtain satisfactory results for images contaminated with salt and pepper noise. Based on the energy functions of the original CV model and the LBF model, a new energy fitting term is constructed to enhance the noise resistance to Gao Si noise and salt and pepper noise. Using the newly constructed CV model, the rough segmentation contour is obtained by using the global information of the image, the rough segmentation contour is taken as the zero level set of the new LBF model, and the accurate segmentation result of the image is obtained by using the local information of the image. At the same time, a new edge detection operator is proposed, which redefines the edge stop function to further improve the noise resistance of the model. Compared with CV and LBF models, combined with Wang and Qi models with global and local information, the proposed model can obtain better image segmentation results and has strong anti-noise performance.
【作者单位】：喀什大学计算机科学与技术学院;上海海洋大学信息学院;
【基金】：国家自然科学基金资助项目(61561027) 国家教育部青年专项资助项目(ECA150375) 新疆高校科研计划青年资助项目(XJEDU2016S076)
【分类号】：TP391.41

【相似文献】