Syntactic Shape Recognition Using Attributed Grammars.

The method is extended to recognize partially distorted shapes and shows the advantage of the syntactic approach using attributed grammars over other existing shape recognition methods.

Abstract

A syntactic approach is applied to the shape description and recognition. The structure of a shape is described by grammatical rules and the local details by primitives. Four attributes are proposed to describe an open curve segment, and the angle between two consecutive curve segments is used to describe the connection. The property of the attributes and the recognition capability of this method are studies. The primitive extraction and syntax analysis can be performed in the same step by using both semantic and syntactic information, namely, the attributes and production rules. The recognition system was implemented and tested on the recognition of airplane shapes. The performance is quite satisfactory with respect to accuracy and computational efficiency. The method is extended to recognize partially distorted shapes. The distorted portion of the shape can be measured in terms of error-weight. The class membership functions of different shapes and the error-weight estimation of the distorted portions are included in the extended recognition algorithms for recognizing noisy and distorted shape patterns. The success of this extension shows the advantage of the syntactic approach using attributed grammars over other existing shape recognition methods.