Tide Table Digit Recognition Based on Wavelet-Grid Feature Extraction and Support Vector Machine

Abstract

To be represented in tabular form and graphical format in ship electronic navigation system, printing tidal material must be processed into textual information, which is completed by an automatic tide table recognition module consisting of a feature extractor and a classifier. In feature extraction, a new wavelet part grid feature is defined based on wavelet's directive characteristics. In classification phase, multi-class SVM classifier is used instead of neural networks. Experiments show that the wavelet grid feature has good stability and satisfactory distinction, and SVM classifiers have better generalization performance than that of neural networks.