基于改进CV模型的金相图像分割

发布时间：2018-06-04 20:38

本文选题：金相图像分割 + 晶粒评级　；参考：《工程科学学报》2017年12期

【摘要】：对金相图像进行快速精确分割是金相晶粒评级的关键步骤,利用传统Chan-Vese(CV)模型很难将晶粒精确地提取出来.为了更加精确地对金相图像进行分割,提出一种基于改进CV模型的金相图像分割方法.初始化水平集函数,对曲线内外两部分分别计算其倒数坎贝拉距离,并将该距离的大小作为拟合中心的权重系数,有效抑制了噪声点对区域拟合中心准确性的影响;引入指数熵自适应调节曲线内外能量权重,减少固定能量权重对曲线演化的影响;同时加入距离规范项以避免水平集函数的重新初始化,加速该模型的收敛.实验结果表明,与传统CV模型、测地线活动轮廓模型、距离规范项的水平集模型以及偏置场修正水平集模型相比,所提方法分割出的金相图像更加精确,分割效率较高且模型收敛性较好.
[Abstract]:The fast and accurate segmentation of metallographic images is a key step in the grading of metallographic grains. It is difficult to extract the grains accurately by using the traditional Chan-Veseau CVV model. In order to segment metallographic image more accurately, a method of metallographic image segmentation based on improved CV model is proposed. Initialize the level set function, calculate the reciprocal Cambera distance for both the inside and outside parts of the curve, and take the distance as the weight coefficient of the fitting center, effectively restrain the influence of the noise point on the accuracy of the region fitting center. The exponential entropy adaptively adjusts the energy weight inside and outside the curve to reduce the influence of the fixed energy weight on the evolution of the curve, and the distance gauge term is added to avoid the reinitialization of the level set function and to accelerate the convergence of the model. The experimental results show that the proposed method is more accurate than the traditional CV model, the geodesic active contour model, the horizontal set model of the distance specification term and the offset field modified level set model. The segmentation efficiency is higher and the model convergence is better.
【作者单位】：南京航空航天大学电子信息工程学院;北京科技大学新金属材料国家重点实验室;
【基金】：国家自然科学基金资助项目(61573183) 新金属材料国家重点实验室开放基金资助项目(2014-Z07)
【分类号】：TG115.21

【相似文献】