Integrating graphics and vision for object recognition (Q2739562)

This book develops a so-called RMR (Render, Match, Refine) approach based on an idea of Besl and Jain (86) mixing computer graphics synthesis and image analysis in order to provide a refinement loop for image interpretation.NEWLINENEWLINENEWLINEAfter an introduction and a state-of-the-art survey in object recognition, chapter 3 describes the graphical rendering methods used to predict a scene aspect. Chapter 4 presents the comparisons processes that can be used to measure the similarity of the prediction with the real image. Now that we have prediction and comparison, it is possible (in chapter 5) to perform iteration in order to converge to an optimal solution according to the comparison criteria. Various algorithm are compared (classical and genetic ones). Chapter 6 discuss about experiments an provide statistical analysis of the overall algorithm applied to different scenes. Conclusion and future works follow.NEWLINENEWLINENEWLINEIt can be added that all the material of general interest used is quite deeply described and explained from the low level, allowing newcomers to get insight in these methods through an interesting application.