Visual Measurement And Electromagnetic Wave Simulation Of Grinding Surface Roughness Based On Color Images

As an important part of surface texture parameters,surface roughness value is closely related to the matching properties,wear resistance,fatigue strength,contact stiffness,vibration and noise of a mechanical part.And it is important for the service life and reliability of mechanical products.In addition,surface roughness also has a great influence on the optical properties,biocompatibility,and friction properties of mechanical parts.Therefore,accurate,efficient,and reliable measurement of surface roughness values is of great significance to the manufacturing industry and the current intelligent manufacturing industry.Due to that machine vision technology has the advantages of high measurement efficiency,low cost,non-contact mode and online support,it has attracted the attention of many scholars and has been applied to the field of roughness measurement.However,the current machine vision-based roughness measurement method still has some problems that need to be solved,including the design and evaluation of image-based feature indices,experimental optimization,and numerical reproduction of the mechanism.In response to these problems,this article takes grinding surface as an example,and makes corresponding research and exploration work:（1）In current machine vision-based roughness measurement methods,the performance assessment of image-based indices associated with roughness mainly focuses on measurement accuracy but ignoring the monotonicity,stability and efficiency of the indices,and most of the indices are designed based on gray image but ignoring the multidimensional nature of color image.To address the problem,the performance assessment indices of video quality prediction model designed by Video Quality Expert Group（VQEG）are used to assess the performance of image-based indices associated with roughness.Based on the difference in area of the diffusion region between the virtual images formed by a light source on surfaces with different roughness levels,energy difference（ED）based on color image is proposed to measurement roughness.The experimental results show that the index performance assessment method can better characterize the index performance.Compared with indices（G_a、RR、F₃）based on gray image,the designed index ED has better comprehensive performance.（2）The latest research findings have proved that surface roughness measurement method based on color images has remarkable superiority.This is because color image can better objectively characterize rough surface due to that it can carry more information compared with gray image.But there are still some problems remained to be analyzed,including the influences of the light source and roughness ranges on the roughness evaluation ability of the Human Vision System（HVS）,the influences of the light source and roughness ranges on the roughness measurement performance of image-based feature indices.To overcome these problems,this paper captures a series of images of 42 grinding specimens under different experimental conditions.At first,the influences of color blocks（red-green（RG）color block,red-blue（RB）color block and blue-green（BG）color block）,brightness level and roughness ranges on the roughness evaluation ability of HVS are analyzed.Then,their influences on the roughness measurement performance of color distribution statistical matrix（CDSM）-based feature indices are analyzed.The main results are as follows:the visual effect of HVS-based method and the roughness measurement performance of the CDSM-based feature indices can be improved by using the RB color block;the resolution of HVS-based surface roughness evaluation method decreases with the increase of surface roughness;the energy index E_p can be used to measure grinding surface whose 0μm（＜）R_a≤0.1μm（100nm）with a mean absolute error（MAE）of 10.5nm;even under the condition of low energy field,the selected feature indices can also used to measure grinding surface roughness with high accuracy and high stability,which is of great significance for energy conservation.（3）There are many types of workpieces in actual engineering,and the imaging environment is complex and diverse.To apply machine vision measurement technology to actual production,there are many experimental factors need to be considered,such as workpiece material,workpiece texture direction,and environmental illumination.How to set the imaging experiment conditions for some specific metal material grinding workpieces is a very complicated problem,and it is extremely time-consuming to use traditional orthogonal experiments for analysis.To address this problem,surface morphology characteristics grinding samples with four different materials（45#steel,iron（Fe）,aluminum（Al）,and copper（Cu））are studied by comparing roughness results obtained by stylus profilometer and white light interference（WLI）method.The comparison results can provide some technical guidances for roughness accurate calibration for workpieces with different materials.Secondly,Taguchi’s experimental method is used to design the imaging experiment for the grinding samples with different materials.The influence of the brightness level of light source,color block,placement of texture direction and environmental illumination on roughness prediction accuracy are analyzed.Finally,according to the analysis results of Taguchi experiment,the best measurement conditions for four type materials under each roughness prediction model are obtained.（4）Although Taguchi method can reduce experimental cost to a certain extent,it is very time-consuming and labor-consuming to analyze all the roughness visual measurement parameters by actual experiments.Moreover,some imaging conditions（such as the incident angle and wavelength of the light source）cannot be accurately and quantitatively controlled in actual experiments.In the present work,we proposed to conduct Electromagnetic Wave Scattering Simulation considering these variables using finite element method（FEM）in order to solove these problems.The simulation process is more theoretically robust,more efficient and cost-effective rather than actual experiments with empirical models.In the simulation process,a simulation model is designed first.In this model,the light incident angles include15°,30°,45°,60°and 75°;the light wavelengths include 400nm,550nm and 700nm corresponding to red,green and blue;the workpiece materials include stainless steel（SUS）,copper（Cu）,aluminium（Al）,and silicon carbide（Si C）.Then,a series of rough surface with different surface roughness values are obtained by defining cosine functions and non-Gaussian surfaces.The simulation results reveal that the machine vision-based roughness measurement method has a certain theoretical measurement range due to the influence of secondary or multiple light scattering.The simulation results can provide certain guidance for roughness visual measurement.