Optical Inspection of Industrial Product

Abstract

The implementation of algorithms based on computer vision to the industry has brought great advantages when carrying out inspections of industrial products, that is why this document focuses on the realization of a 3D inspection of an object through the method of Photometric Stereo. A method that is based on taking images under different lighting directions, thus allowing estimating the map of the surface of the object under inspection. To develop this method, a light ring from the Smart Light company with its respective controller and a Basler monochrome camera was implemented. During the development of this master's thesis, it was possible to observe the operation of each of the stages to carry out the inspection, in this case, the capture of images, the reconstruction method, and the calibration process to convert pixels to real units and thus to be able to compare the dimensions obtained by the algorithm concerning the real measurements of the object. Finally, the response of the system to objects that differ from one another concerning geometry, size, or material is analyzed and thus the precision of the approach used is verified.