Use of laser range finders and range image analysis in automated assembly tasks

A proposition to study the effect of filtering processes on range images and to evaluate the performance of two different laser range mappers is made. Median filtering was utilized to remove noise from the range images. First and second order derivatives are then utilized to locate the similarities and dissimilarities between the processed and the original images. Range depth information is converted into spatial coordinates, and a set of coefficients which describe 3-D objects is generated using the algorithm developed in the second phase of this research. Range images of spheres and cylinders are used for experimental purposes. An algorithm was developed to compare the performance of two different laser range mappers based upon the range depth information of surfaces generated by each of the mappers. Furthermore, an approach based on 2-D analytic geometry is also proposed which serves as a basis for the recognition of regular 3-D geometric objects