文献类型：期刊文献

中文题名：一种用于纤维增强复合材料微观结构图像的阈值分割算法

英文题名：A Threshold Segmenting Algorithm with Application in Fiber Reinforced Composite Microstructure Image

作者：万永菁[1];林家骏[1]

机构：[1]华东理工大学信息科学与工程学院,上海200237

年份：2008

卷号：34

期号：2

起止页码：283

中文期刊名：华东理工大学学报（自然科学版）

外文期刊名：Journal of East China University of Science and Technology

收录：CSTPCD;;Scopus;北大核心:【北大核心2004】；CSCD:【CSCD2011_2012】；

基金：华东理工大学青年骨干教师基金项目(2007-03)

语种：中文

中文关键词：阈值分割;信息熵;纤维增强复合材料;Otsu方法

外文关键词：threshold segmentation; information entropy; fiber reinforced composite; Otsu approach

摘要：提出了一种新的阈值分割方法,将原始图像划分为子图像,计算子图像变化统计分布矩阵及平均加权信息熵能量占总信息熵能量和的比例,判断子图像内是否含有目标纤维。提出一种改进的Otsu阈值分割算法对含有目标的子图像进行分割。实验证明:该方法适用于目标稀疏且对比度低、亮度分布不均匀的图像,应用在纤维增强复合材料微观结构图像阈值分割中,能较好抑制图像中的噪声,在不破坏纤维完整性的同时获得较满意的分割结果。
A novel threshold segmenting approach is proposed. The original image is divided into several sub-images. The statistic distributing matrix of change information in sub-image and the ratio of average weighting information entropy energy to the summation of information entropy are calculated. Whether there is the object fiber in the sub-image is judged and an improving Otsu threshold segmenting algorithm is adopted to segment the sub-image with fiber. The proposed approach can be applied in the segmentation of the image with the characteristic of lower ratio of object to background, lower contrast or asymmetrical lightness distributing. The proposed approach is used in the threshold segmentation of the fiber reinforced composite micro-structure image, the noise in the image is weakened and the satisfying segmentation results is obtained without breaking the fiber.