Generating 3D product design models in real-time using hand motion and gesture

Abstract

Three dimensional product design models are widely used in conceptual design and in the early stage of prototyping during the design processes. A product design specification often demands a substantial amount of 3D models to be constructed within a short period of time. Current methods begin with designers sketching product concepts in 2D using pencil and paper, which in turn are then translated into 3D models by a design individual with CAD expertise, using a 3D modelling software package such as Pro Engineer, Solid Works, Auto CAD etc. Several novel methods have been used to incorporate hand motion as a way of interacting with computers. There are three main types of technology available to capture motion data, capable of translating human motion into numeric data which can be read by a computer system. The first being, hand gesture glove-based systems such as “Cyberglove”, these systems are generally used to capture hand gesture and joint angle information. The second is full body motion capture systems, optical and non-optical-based, and finally vision based gesture recognition systems which capture full degree of - freedom (DOF) hand motion estimation. There has yet to be a method using any of the above mentioned input devices to rapidly produce 3D product design models in real time, using hand motion and gestures. In this research, a novel method is presented, using a motion capture system to capture hand gestures and motion in real time, to recreate 3D curves and surfaces, which can be translated into 3D product design models. The main aim of this research is to develop a hand motion and gesture-based rapid 3D product modelling method, allowing designers to interactively sketch out 3D concepts in real time using a virtual workspace. A database of a number of hand signs was built for both architectural hand signs (preliminary study) and Product Design hand signs. A marker set model with a total of eight markers (five on the left hand and three on right hand/marker pen) was designed and used in the capture of hand gestures with the use of an Optical Motion Capture System. A preliminary testing session was successfully completed to determine whether the Motion Capture system would be suitable for a real-time application, by effectively modelling a train station in an offline state using hand motion and gesture. An OpenGL software application was programmed using C++ and the Microsoft Foundation Classes which was used to communicate and pass information of captured motion from the EVaRT system to the user.