A Surface Roughness Measurement Method Based on Color Image Euclidean Distance

Abstract

Vision-based measurement methods has the advantage of being fast and supporting online measurements, this paper put forward to detect grinding surface roughness by an evaluation method that is euclidean distance of color images. The surface roughness images with different roughness levels are analyzed using a quaternion matrix; the singular value decomposition is carried out for them. The euclidean distance between the singular value of reference image and distortion image is established as the evaluation basis to the surface roughness. The proposed index was compared with the color difference index and verified using support vector machine regression model. The experimental results show that the proposed measurement method exhibits higher accuracy and wider measurement range, showing a good corresponding relationship between quaternion and color vector information, which will provide convenience for automatic measurement. The proposed surface roughness measurement method has a certain potential in engineering application.