Object identification based on 3-D features using structured light technique

Three dimensional (3-D) vision techniques in the field of Computer Vision aims mainly at reconstructing a scene to find its three dimensional geometrical information. Active methods of sensing in a generic sense have better sensing abilities than passive methods since they release some form of energy into the environment for sensing. Structured Light technique in 3-D computer vision is one such active method which throws a light pattern on to the scene whose image is then captured by an intensity camera. The structured light technique is by far the most rugged of all 3-D reconstruction methods in the current scenario. The technique is so rugged that even passive methods such as stereo vision makes use of the pattern projection advantage thus resulting in many hybrid approaches. The paper presents an extremely low cost structured light method for scene reconstruction using off-the-shelf hardware. The objects considered are deliberately made texture-less and no contrast between the object and background is ensured so as to demonstrate the abilities of the structured light technique. The reconstructed information of the object is utilized for identifying its identity. The need comes in the form that such identification processes cannot otherwise be done using 2-D visual features. The concept is validated for a set of three different cylindrical objects which basically vary in their slope of the top surfaces. The paper intends to demonstrate the ease of developing a complicated approach for 3-D vision using simple low-cost hardware and finally appreciates the wide range of application possible.