基于改进GLCM和Tamura特征的工件表面粗糙度检测

Workpiece surface roughness detection based on improved GLCM and Tamura features

由于使用图像法获得工件的单一纹理无法精确描述工件表面粗糙度,所以给出了通过改进GLCM(灰度共生矩阵)特征和Tamura特征的纹理提取算法,并通过建立纹理特征与表面粗糙度之间的模型关系定性评价工件表面粗糙度。首先对传统的GLCM加以修改,提出了增强的GLCM(ET-GLCM)算子,提升了GLCM的尺度不变性和旋转不变性,并经过对比实验证明ET-GLCM有较强的鲁棒特性;然后创建了支持向量机检测模型,测量不同粗糙度等级的工件表面图像。以工件表面图像的纹理特征参数为输入,对应的工件表面粗糙度Ra为期望输出,从而完成工件表面粗糙度的检测。通过实验验证支持向量机检测模型的有效性,其检测结果相对误差不超过5%,绝对误差小于0.06。根据仿真对比实验结果表明,提出的方法具有较高的检测精度,可用于工件表面的粗糙度检测。

Since the surface roughness of the workpiece cannot be accurately described by obtaining a single texture of the workpiece by means of the image method,the texture extraction algorithm of improved GLCM(grayscale co⁃occurrence matrix)features and Tamura features is given,and the surface roughness of the workpiece is qualitatively evaluated by establishing the model relationship between the texture features and the surface roughness.The traditional GLCM is modified,and an enhanced GLCM(ET⁃GLCM)operator is proposed,which improves the scale invariance and rotation invariance of GLCM.The comparative experiment proves that ET⁃GLCM has strong robust characteristics.A support vector machine detection model was then created to measure the surface image of the workpiece at different roughness levels.The texture characteristic parameter of the workpiece surface image is taken as the input,and the corresponding workpiece surface roughness Ra is the expected output,so as to complete the detection of the workpiece surface roughness.The effectiveness of the support vector machine detection model is verified by experiments,and the relative error of its detection results does not exceed 5%,and the absolute error is less than 0.06.The results of the simulation comparison show that the proposed method has high detection accuracy and can be used for the roughness detection of the workpiece surface.