Syntactic pattern recognition oriented large scale settlement matching algorithm

Settlement matching is one of the important parts of large scale map update. All edges of settlements are classified based on their length and orientation so as to translate them into set of symbols at first. The representative patterns of settlements were thoroughly enumerated and their corresponding grammars are constructed based on common settlement types secondly. Thirdly, all settlements are classified into four types i.e. rectangle type, crenel type, `L´ type and irregular type supported with syntactic pattern recognition algorithm assisted by Artificial Neural Network. And different algorithms responding to different settlement types were developed finally. Test illustrates that this method is more advanced in classifying the settlement, is more effective to reduce the complexity, and run faster in the process of settlement matching.