Grinding surface roughness detection method based on quaternion singular value entropy index

Abstract

The invention provides a grinding surface roughness detection method based on a quaternion singular value entropy index, and belongs to the technical field of roughness detection. The grinding surfaceroughness detection method includes the following steps: using three imaginary parts that are pure quaternions to respectively replace three color components of a color image; carrying out singular value decomposition on a pure quaternion matrix; intercepting previous 30 singular values, and calculating a singular value entropy index; and carrying out training fit on the singular value entropy index and a roughness value by using a support vector machine, thereby predicting a roughness. The method of the invention processes color information as a whole, gives a reasonable data structure representation, and may directly obtain a picture corresponding to a virtual image irradiated by a color light source on a surface of a to-be-detected object to detect the roughness; and as the image is the color image rather than a gray image, an image degradation process does not exist, and the accuracy of image processing is improved; and therefore, the accuracy of a roughness detection result is higher.

Description

【Technical field】 [0001] The invention relates to the technical field of workpiece surface roughness detection, in particular to a grinding surface roughness detection method based on a quaternion singular value entropy index. 【Background technique】 [0002] The machine vision detection method of surface roughness generally analyzes the grayscale image, objectively quantifies the grayscale value information of the image, and then uses image texture analysis technology (image texture analysis technology is generally classified into three categories: frequency spectrum, structure and statistics), Get the relationship between surface roughness and machine vision index. Since the grayscale image is a degraded image, the sensitivity of image features to surface roughness is reduced to a certain extent, and it cannot be judged subjectively well visually; moreover, the application of more grayscale co-occurrence matrix methods requires Combined with a microscopic device, the test ...

AI Technical Summary

Problems solved by technology

Since the ground sample itself does not have the characteristics of distinct colors, it is difficult to obtain significant color difference information by directly photographing the sample

At present, there is no related report on the use of singular value entropy index to characterize surface roughness

Images