【摘要】 在磨粒自动识别系统中,首先对彩色磨粒图像进行预处理,运用图像增强、自适应阈值选取、磨粒的标识和图像二值化方法成功地提取了特征磨粒;然后根据磨粒的识别特征建立描述磨粒形态的特征参数体系,确定了磨粒的三类特征参数(颜色、表面纹理和形状尺寸参数),并对磨粒进行特征量提取创建参数数据库;最后以提取的磨粒特征量为基础,运用灰色定权聚类的方法成功地识别了6种特征磨粒(正常磨粒、球形磨粒、切削磨粒、严重滑动磨粒、Fe2O3磨粒、Fe3O4磨粒).实验表明所提方法切实可行.更多还原

【Abstract】 In the system of automatic identification for wear particles,the pretreatment methods including image enhancement,selection of adaptive threshold,marking debris and binary image are successfully applied to extract debris information from the color debris image.Then the characteristic parameter system of wear particles is produced by studying recognition character of wear particles.Three types of characteristic parameters（color,surface texture and shape and size） are confirmed.And a parameter database is founded with the extracted numerical descriptors of debris.Finally,the theory of fixed-weight-grey-clustering is successfully used to classify six types of wear particles（rubbing,spherical,cutting,severe sliding,Fe₂O₃ and Fe₃O₄ wear particles） according to their morphology numerical parameters analysis.The experimental results show that the methods proposed are feasible and effective.更多还原