Image analysis using new Descriptors Average Feature Optimization based on Fourier Descriptors technique

Abstract

Fourier descriptors technique has been used to extract the shape of an image based on Fourier analysis of its boundary. In this paper, a new feature extractor technique named Descriptors Average Feature Optimization (DAFO) based Fourier Descriptors shape descriptor is introduced. In order to examine the ability of this technique, this feature extractor is evaluated based on different image sizes and rotations and thus compared with Fourier descriptor in the same order of features. Intra-class analysis is a set of equations that has been implemented based on Total Percentage Min Absolute Error (TPMAE) with image Rotation Scale Translation (RTS) in order to measure the performance of the new DAFO technique. The analysis results indicate that the new DAFO technique is able to produce the small values of TPMAE as compared to normal Fourier descriptors.