摘要
为了保证贝氏体组织识别的准确度,提出了一种基于平面几何和梯度加掩模的分水岭相结合的贝氏体分割技术。首先应用平面几何算法对梯度图像进行预处理。通过考察梯度图像中各像素点与其附近采样点间曲率变化情况,来判断考察点是否为贝氏体组织,滤除组织中的非贝氏体。用梯度加掩模的分水岭方法对预处理后的梯度图像进行分割。实验结果证明,提出的算法有效抑制了非贝氏体组织的影响,避免了过分割,提高了贝氏体组织的计算机自动识别的准确性。
In order to ensure the accuracy of bainitic microstructure identification,a novel image segmentation method based on plane geometry and watershed algorithm is proposed.First the gradient image is preproeessed according to plane geometry algorithm.By examining the curvature change between the image pixel and it＇s neighbor sampling point,estimate the image pixel is the bainite or not.According to this,the nonbaintic microstructure can be removed.Finally,the preprocessed gradient magnitude image is segmented by the watershed transform.Experimental results indicate that this algorithm can effectively inhibit the impact of non-bainitic microstructure,avoid over-segmentation,improve the accuracy of automatic identification.
出处
《计算机工程与应用》
CSCD
北大核心
2009年第27期166-167,170,共3页
Computer Engineering and Applications
基金
河北省教育厅科研计划项目No.2006143~~
关键词
图像预处理
曲率
距离
分水岭变换
image preprocessing
curvature
distance
watershed
