Image retrieval based on coefficient correlation index

Abstract EN

Image retrieval is an active research area in image processing, pattern recognition, and computer vision.

In this proposed method, there are two techniques to extract the feature vector, the first one is applying the transformed algorithm on the whole image and the second is to divide the image into four blocks and then applying the transform algorithm on each part of the image.

In each technique there are three transform algorithm that have been applied (DCT, Walsh Transform, and Kekre’s Wavelet Transform) then finding the similarity and indexing the images, useing the correlation between feature vector of the query image and images in database.

The retrieved method depends on higher indexing number.

Experimental results have shown better results (higher precision and recall) by applying DCT on the image than the other transform algorithms and the performance improvement if dividing the image into equal four blocks and applying the transformed algorithm into each part.