A Comparative Analysis of Various Local Feature Descriptors in Content-Based Image Retrieval System

Abstract

Image acquisitions are increasing day by day due to progress in social networking and digital technologies. Nowadays, with the evolution of various image capturing devices, an enormous quantity of complex images is being produced.content-primarily based image Retrieval (CBIR) is the answer to access images without difficulty wherein proper indexing and association are required. It makes CBIR a distinguished field in computer vision research. There are several uses of CBIR systems in day to day life for example medical, internet, scientific research and various other communication media. In the CBIR system, the user gives a query to obtain images from large datasets having a large number of images. In information transfer via electronic media using particular formats of data, images play an essential role. The extraction of information from communicated images is necessary with extra processing. In this paper, a comparative survey has been carried out on different content-based image retrieval implementation. These methods are determined by several authors for the feature extraction process of images and classification. This will help to plan the strategy for optimizing the CBIR system.

Unleashing the Power of Evolutionary Machine Learning in CBIR: A Study of Kernelized Radial Basis Auto-Encoder Function Neural Network

“…One of the main research problems that the multimedia community has been studying for decades is CBIR (Content-Based Image Retrieval) (Tiwari & Pant, 2022). The core of CBIR is image representation, as it aims to find images by analyzing their visual contents (Tyagi et al 2021). Global features such as color features, edge features, texture features, GIST, and CENTRIST are commonly used.…”