<XML><RECORDS><RECORD><REFERENCE_TYPE>10</REFERENCE_TYPE><REFNUM>8489</REFNUM><AUTHORS><AUTHOR>Siebert,J.P.</AUTHOR></AUTHORS><YEAR>1987</YEAR><TITLE>Vehicle Recognition Using Rule Based Methods</TITLE><PLACE_PUBLISHED>Turing Institute Research Memorandum, TIRM 87-18, 1987, Turing Institute, Glasgow, Scotland.</PLACE_PUBLISHED><PUBLISHER>N/A</PUBLISHER><LABEL>Siebert:1987:8489</LABEL><KEYWORDS><KEYWORD>pattern recognition</KEYWORD></KEYWORDS<ABSTRACT>A method of distinguishing 3D objects within a 2D image by application of an ensemble of shape feature extractors to the 2D silhouettes of the objects is presented. Measures of shape features extracted from example silhouettes of objects to be discriminated are used to generate a classification rule tree by means of computer induction. This object recognition strategy was successfuly used to discriminate between silhouettes of model cars, vans and buses viewed from a constrained elevation but all angles of rotation. The rule tree classification performance compared favourably to MDC (Minimum Distance Classifier) and KNN (K Nearest Neighbours) statistical classifiers in terms of both error rate and computational efficiency. An investigation of these rule trees generated by example has indicated that the tree structure is heavily influenced by the orientation of the objects, and groups similar object views into single decisions. It has been further demonstrated that the system is relatively insensitive to small variations in camera elevation and also continued to classify correctly when a thermal imager was substituted for the visual camera used to train the system.</ABSTRACT><NOTES>Vehicle dataset available at: http://www.ailab.si/orange/datasets/vehicle.htm</NOTES></RECORD></RECORDS></XML>