An Assessment Method of the Data Similarity Considering the Shape Differences of Objects

Abstract

The measure of similarity is an essential basic work in the field of pattern recognition, machine learning and data mining, etc. The shape similarity distance (SSD) which considering vector difference factors in the calculation of similarity, is an amendment of the classical distance. Based on the classical Euclidean distance, Manhattan distance and the SSD, this paper completed the clustering on the multiple datasets that contained shape features data of objects. The experimental results indicated that compared with the classical distances, the SSD can consider the characteristics of shape of theses objects to estimate their similarity.