The Application Of Machine Vision Non-Lambertian Surface Imaging Theory In Surface Detection

When using the visual image technology to detect surface defects of surface feature parts,a complex curved surface image including the defective area is formed by the effect of poor imaging consistency.Aiming at the problem of surface defect extraction in complex surface,the non-Lambert surface imaging theory and surface defect detection method are studied.Firstly,aiming at the geometricl optics model of directional reflection and hemispherical emission,a parallel light source which is easy to control and model analysis is adopted.The bidirectional reflectance distribution function is decomposed into linear combinations of four basic functions based on the thought of surface reflection light decomposition,and the general expression of plane's reflection model is obtained.Three representative surface radiation models belonging to the general combination model are studied and analyzed emphatically,including Lambert reflection model,Oren-Nayar reflection model and Ward reflection model.Three reflection models' reflection maps are obtained by using numerical calculation.Then,the space radiation distribution of complex surface is obtained by combining plane radiation model with surface equation.The O-shaped surface theoretical radiation intensity values on different illumination and observation directions are calculated by using the above three kinds of radiation models.The O-shaped rubber sealing ring is used as the test sample,and the radiation measurement experiment on its local and entire upper surface is carried out.The experimental results show that the rubber sealing ring surface is more in line with the Ward reflection model.The characteristic parameters of the high brightness area on curved surface are used as the reference,the difference between the measured and simulated images are quantitatively described.The parameters in the Ward reflection model are estimated based on the idea of minimizing variance.The orientation of local surface on O-shaped ring is analyzed by combining the contour shape change of high brightness area on the measured image with the simulated image.The number of high brightness areas in the surface image with surface defects is not equal to the number of high intensity areas in the surface simulation image without surface defects.All the high brightness areas in the curved surface image are extracted according to the gray level threshold determined by the simulation image,and the defective region on the curved surface is obtained by combining with the pixel area constraint of high brightness region on the normal curved surface.The experimental results show that the method based on the extraction and analysis of high brightness areas on the surface can extract the defects such as dents and flashes on the O-shaped ring suface,and give the occupied pixel area.It is difficult to detect all the surface defects from a single image,but it is possible to reduce the number of images collected without losing the surface defect information by selecting the illumination direction of the light source and the camera observation orientation.