Using image processing based on neural networks in reverse engineering

Various techniques have been employed to recover three-dimensional (3D) sur faces in the redesign of products, customized designs, and the building of virtual environments. Reverse engineering methodology provides an efficient tool for the manufacturing of free-form and sculptured shapes. This paper describes an image processing approach to the 3D shape recovery based on ne ural networks that tackles the major bottleneck in the current reverse engi neering process, namely the lack of a rapid Link between the physical objec t and its design representation. A range of applications has shown this met hod to be feasible and efficient. (C) 2001 Elsevier Science Ltd. All rights reserved.