Content Based Image Retrieval using Heron Filtering, Fast BEMD and Gray level features

In today’s world, large amount of digital image data is generated and stored in the repository. Finding for a particular image/s accurately in the repository poses the challenge. The aim of the proposed work is to search for relevant images from the repository based on user query image. In this work, the low-level features of images present in themselves such as edges and gray levels are used to retrieve the similar images. A novel approach is proposed in this paper by using a new filter applied on images called Heron mean filter, to smoothen the image, then combination of the edge features extracted by Fast and Adaptive Bidimensional Empirical Mode Decomposition (FABEMD) technique and gray level features through histogram. The experiments are conducted using Wang’s dataset of 10 categories of 100 images each. The tabulated results are obtained by using different similarity measures like Euclidean distance and Bhattacharya’s Co-efficient method. It is observed from the results that there is an improvement in the proposed system in terms of average precision and average recall, against the existing system.