Review Of Feature Extraction Techniques In Content Based Image Retrieval

Abstract

Today, image data exists with extremely diverse visual and semantic content, and it is rapidly growing. This created innumerable possibilities and hence considerations for real-world image search system designers. Manual annotation of images with keywords describing the image content can make it easier to find images of interest, but this takes more time, making this approach very costly [1,2]. Thus, searching in image collection based on visual content is potentially a very powerful technique. CBIR uses visual contents of an image to search the desired images. It deals with the fundamental problem of mathematically describing the image content (image signature) and then, assessing the similarity between a pair of images based on their signatures. Despite the apparent simplicity of this, there are significant obstacles that need to be overcome to design an efficient CBIR system. It is important to find a good image signature using suitable image features for the design of an efficient CBIR system. Most of the CBIR approaches rely on a preprocessing step of feature extraction, which aims to extract suitable image features (descriptors) such as color, texture, shape, and spatial layout, that carries enough information to allow successful retrieval of relevant images from a database containing thousands or millions of images.