Ship Outline Feature Selection Using B-Spline Function

摘要

This thesis presents two methods of ship classification with IR images. The first method is the Fourier Coefficient method which transform the sample points of a superstructure profile of a ship to the spatial frequency components. The shape of the coefficient curve can be used to classify the type of a ship from 8 different categories. But, the differences is so minor that it is difficult to implement a computer program to recognize it. The second method is a B-spline Coefficient method which uses the uneven spaced spline coefficients to find the beginning, the peak, and the area of the lumps of a ship for classification. This method is better than the Fourier Coefficient method. The study of these methods is presented here.