Significantly improving scan-based shape representations using rotational key feature points

Abstract

In a previous paper we have presented the idea of representing the shape of a 2D object by scanning it following a Hilbert curve then performing wavelet smoothing and sampling. We also introduced the idea of using only a subset of the resulting signature for comparison purposes. We called that set the Key Feature Points (KFPs). In this paper we introduce the idea of taking the KFPs over a number of views of the original shape. The proposed improvement results in a significant increase in recognition rates when applied to the MPEG-7 and ETH-80 data sets when the Hilbert scan is used. Similar improvement is achieved when the raster scan is used. © 2010 Springer-Verlag.