基于改进测地线轮廓模型的图像分割算法  

Image Segmentation Algorithm Based on an Improved Geodesic Contour Model

作　　者：许亚军[1] 韩应征[1] 

机构地区：[1]太原理工大学信息工程学院,太原030024

出　　处：《科学技术与工程》2014年第4期241-244,共4页Science Technology and Engineering

基　　金：山西省自然科学基金(20130110173)资助

摘　　要：针对测地线轮廓模型对初始轮廓的敏感度高的情况,提出一种改进的测地线轮廓模型。该模型在测地线轮廓模型的基础上,引入区域方差概念,加快了初始轮廓的演化效果;同时,在几何化模型的能量函数中加入判别函数,进一步提高了分割速度与目标背景区分度。仿真实验结果表明,改进测地轮廓模型较原模型对初始轮廓的放置有很好的鲁棒性,是一种有效的图像分割算法。Aiming at the geodesic contour model to the sensitivity of the initial contour is high, an improved ge- odesic contour model is put forward, this model using regional variance concept, speeding up the initial contour evolution effect is based on the geodesic contour model, at the same time in geometrization model of energy function, join discriminant function, further improve the speed of segmentation and target differentiate background. The simulation results show that the improved geodesic contour model has a good robustness, and it is better than the original model in the placement of initial contour.

关 键 词：区域方差 判别函数 初始轮廓 鲁棒性 

分 类 号：TN911.73[电子电信—通信与信息系统]