ICTACT Journal on Image and Video Processing

V. Balamurugan ¹, K. Senthamarai Kannan ², S. Selvakumar ²

Affiliations
1 Department of Computer Science and Engineering, Sri Vidya College of Engineering and Technology, Tamil Nadu, IN
2 Department of Statistics, Manonmaniam Sundaranar University, Tamil Nadu, IN

Abstract

Searching for similar sequence in large database is an important task in temporal data mining. Similarity search is concerned with efficiently locating subsequences or whole sequences in large archives of sequences. It is useful in typical data mining applications and it can be easily extended to image retrieval. In this work, time series similarity analysis that involves dimensionality reduction and clustering is adapted on digital images to find similarity between them. The dimensionality reduced time series is represented as clusters by the use of K-Means clustering and the similarity distance between two images is found by finding the distance between the signatures of their clusters. To quantify the extent of similarity between two sequences, Earth Mover’s Distance (EMD) is used. From the experiments on different sets of images, it is found that this technique is well suited for measuring the subjective similarity between two images.

Keywords

Similarity Search, Vector Quantization, Similarity Measures, Clustering, EMD.

Full Text