头部MRI图像外轮廓提取算法的实现与比较  

IMPLEMENTATION AND COMPARISON OF OUTER CONTOUR EXTRACTION ALGORITHM OF MRI HEAD IMAGE

作　　者：张欣然[1] 陈琪儒[1] 何煦佳[1] 杨荣骞[1] 

机构地区：[1]华南理工大学生物科学与工程学院,广东广州510006

出　　处：《计算机应用与软件》2013年第5期11-14,共4页Computer Applications and Software

基　　金：国家自然科学基金项目(81101130;20776055);广东省大学生创新实验项目(1056111108)

摘　　要：为精确提取头部MRI图像外部轮廓,提供后期三维重建数据,根据对边缘检测、迭代阈值、自适应阈值法和半阈值分割法在MRI断层图像分割效果的比较,提出一种改进的头部MRI图像轮廓提取方法。该方法在对图像进行预处理后,通过直方图统计结果对图像进行半阈值分割后得到二值图像,由二值图像梯度信息得到边界点集,通过八邻域搜索提取得到包含多条轮廓的轮廓数据集,并根据长度、面积等判定条件准确提取出断层截面的外轮廓。实验结果表明:该方法能提取出更为准确和连续的轮廓,较其他方法拥有更高的成功率。In order to precisely extract outer contour line of MRI head image for post-3D reconstruction data preparing,we propose an improved contour line extraction method for MRI head image according to the comparison of segmentation results of MRI tomographic image using edge detection,iterative threshold,adaptive threshold and semi-threshold segmentation respectively.In this method,we first segment images into binary images based on semi-threshold segmentation through histogram statistics result after the pre-treatment of the image,and derive boundary point set from gradient information of binary image;then we gain the contour lines dataset which contains multiple contours through 8-neibghourhood search,and accurately extract the outer contour line of tomographic cross-section based on the determinant conditions of length and area and so on.Experimental results show that the method can be used to extract more accurate and continuous contour lines and has a higher success rate than other methods.

关 键 词：图像预处理 半阈值分割法 基于梯度的轮廓提取 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]