摘要
针对重轨生产线钢坯检测所涉及的钢坯提取问题,提出了一种改进的交互式图论分割方法.首先用基于颜色差异的改进的K均值聚类算法将人工标记的种子进行精确地初步分类;然后使用改进的图论分割算法,将钢坯目标从复杂场景中分割出来;最后将分割结果进行边缘校正和去噪处理.实验结果表明:该算法充分利用了图像的区域特征和边缘特征,提高了分割的质量和速度,分割结果满足实际应用的需求.
For billet extraction in steel heavy rail production process of billet detection,an improved interactive graph cuts method was proposed.First the labeled seeds were accurately classified using an improved K-means clustering algorithm based on color divergence,and then objects from image were divided by an improved graph cuts algorithm.Finally the edge was corrected and the noise was removed.Experiment results show that this algorithm can make full use of the characteristics of region and boundary of images improving the speed and quality of segmentation,and the segmentation result can meet the need of practical application.
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2012年第7期57-61,共5页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家自然科学基金资助项目(50975211
61175013)
武汉市科技攻关项目(200810321164)
湖北省自然科学基金资助项目(2010CDB11107)
武汉市学科带头人计划资助项目(Z201051730001)
