Document Type
Thesis
Date of Award
9-30-1990
Degree Name
Master of Science in Mechanical Engineering - (M.S.)
Department
Mechanical and Industrial Engineering
First Advisor
Rajesh N. Dave
Second Advisor
Nouri Levy
Third Advisor
Joshua Greenfeld
Abstract
In this report an integrated method of classifying aerial images as related to photo stereo grammetry is discussed. In this method, an investigation of Fourier descriptors as features for the edge matching approach is carried out. The efficient calculation of the Fourier Descriptors is made possible by applying the polygonal approximation method to the original edges. In order for the features to be invariant to rotation, scaling and starting point shift, a normalization technique is adopted. Initially the Fourier Descriptors are tested for synthesized test datasets and it was proved that the variances of the Normalized Fourier Descriptors was a good candidate as a classifier. This was later tested with real edges. In order to check for partial shape matching an additional feature is utilized, this feature is derived from the ratio of the edge lengths. It is shown that the Fourier Descriptors fall into a class of tools that are unique in boundary description.
Recommended Citation
Kittappa, Ananth, "Edge matching of aerial scenery using fourier descriptors" (1990). Theses. 2801.
https://digitalcommons.njit.edu/theses/2801
