Active object recognition using sensor planning strategy

• Main Author: Liu, Tian Rong.
• Other Authors: Chan, Kap Luk
• Format: Theses and Dissertations
• Published: 2010
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Electronic systems
• Online Access: http://hdl.handle.net/10356/39000
• Institution: Nanyang Technological University

In recognizing 3D objects that are symmetric or similar, ambiguity may occur if the image is taken from a single arbitrary position. Hence an active recognition approach is desired. This thesis focuses on such an approach. A framework for recognizing a 3D object from multiple views of the object acquired through a sensor planning strategy is developed in this research. Three major issues are addressed in this research : (1) the representation of reference model and sensed object, (2) the generation of hypotheses set and its update over subsequent sensing for possible match between the sensed and the reference models, and (3) the sensing strategy for planning the next viewpoint of the camera for reliable object recognition.