Paper Details: Downloads: 825
Serial Number: P1151112535
Title: Spatial Domain Approach for Effective Image Retrieval using Thumbnails of Images
Authors: V. R. Khapli and A. S. Bhalchandra
Abstract: The paper discusses the technique for making image retrieval effective using image thumbnails. Here effective means faster and accurate. The proposed technique suggests the use of Vector quantization to extract features from image thumbnails database as well as query image thumbnail. The empirical results shows that the thumbnail size 8 is suitable for faster image retrieval without compromising accuracy, in VQ based image retrieval. It was also tested that Euclidian Distance is not always the best similarity metric for effective image retrieval. Experimentation was performed to find best similarity metric for VQ based image retrieval technique and it was found that Kolmogorov distance outperformed Euclidian Distance in terms of retrieval accuracy and retrieval time.
Keywords: Content based image retrieval, Codebook, Vector quantization, Similarity metrics
Journal/Conference: International Journal of Graphics, Vision and Image Processing
Volume: 11
Issue: 3
Submission Date: 3/19/2011 12:00:00 AM
Review Date: 4/8/2011 12:00:00 AM
Publishing Date: 4/23/2011 12:00:00 AM
Article Downloads: 825
Download:

Facebook