Object Recognition and Sorting By Using a Virtual Cartesian Robot with Artificial Vision

This work presents a system able to recognize the 26 capital letters of the Spanish alphabet. The letters are 5×3cm black-painted objects randomly distributed in different places and orientations over a flat-surface white-painted table. We utilize a Web-cam devices plus VFW video libraries in order to obtain the images corresponding to each letter. Thereafter, in order to be able to identify the letters we used several digital signal processing (DSP) algorithms. Once that the letter-recognition phrase has been finished we command a virtual cartesian robot to reorder the letters so that any arbitrary user pre-selected text may be form and read. First the quality of the image so obtained is improved by digital processing and then we perform a pattern recognition algorithm by using a method that we call digital signature based on a QuadTrees image processing method. In addition to perform the letter recognition phase we estimate the exact position and orientation of the given letter within the table. With this information our virtual robot can proceed to reorder the letter as prescribed by the user. In order to visualize and optimize the movements of the Cartesian robot we develop a 3D simulator based on the libraries GLSCENE OpenGL (Delphi).