This paper describes a method of cavity features and use of neuralnetwork for recognizing handprinted Thai characters, The recognitionprocess is implemented using mathematical morphology to detect thecavity features of patterns, and learning to classify by neural network.The recognition divided into three stages. First, the handprinted Thaicharacters are segmented from the sentence into three different levelgroups. Then, the cavity features of each handprinted Thai character aredetected, and numbers counted by the Euler number method. Finally, thearticle uses the majority area of the cavity features for computing thefeature codes of the characters in each class. These codes are trainedby neural network for learning the classification characters
展开▼