Micro-Vision Image 3D Topography Reconstruction And Roughness Detection Of Workpiece Surface

3D detection and evaluation of workpiece surface topography is a hot spot in surface field in recent years. In allusion to complex equipment and high cost of 3D detection, the surface roughness detection method based on 3D reconstruction of the micro-vision images is proposed in this paper. This paper takes computer micro-vision as the detection means. The 3D topography and roughness parameters of the workpiece surface is obtained by 3D reconstruction of the gray images of the workpiece surface using shape from shading.This paper analyzes the effect of the illuminating direction and imaging parameters on the image quality in imaging process. The experimental system of the acquisition of micro vision image is constructed, including designing illuminant and the demarcating of the camera. Experimental results show that the self-designed LED lighting can provide uniform illumination for the measured object and eliminate the effect of strong specular reflection. The images of workpiece surface with high definition and rich details are obtained.The preprocessing algorithms of the workpiece surface image are studied. The original images are grayed and the texture of image is corrected. The character of gray distribution of surface image and the factors that effect photo-taking are analyzed, which is used to estimate character of noise, and then the noise is filtered out by Wiener filter. The distortion of gray difference caused by light occlusion and strong reflection of the summit is revised. The results showed that each gray value can represent the correct height difference after pretreated, which is the foundation for 3D topography recovery.An improved illuminant model applicable to 3D topography recovery of workpiece surface can be obtained by weighted combination of the simplified Oren-Nayar diffuse reflection model provided with a specular reflection term according to reflective characteristics of the metal micro-surface and gray distribution of images. Moreover, the corresponding algorithm based on the improved illuminant model for minimization approach in SFS method is researched. The conversion model between gray difference and height difference are established in regression method. The iterative initial value which makes convergence faster and higher accuracy of convergence is provided. The 3D reconstruction experiments of turning surface images, boring surface images and end milling surface images are done. The profile curve of reconstruction results is consistent with the results measured by measuring roughness instrument. The error of roughness parameters between the reconstruction and the measurement is about 10%. The results show that the surface roughness can be detected quickly and accurately in this method. The new ideas and methods about the in-situ detection of roughness is provided.