An Integrated Framework for Information Identification With Image Data Using Multi-Technique Feature Extraction

An Integrated Framework for Information Identification With Image Data Using Multi-Technique Feature Extraction

Rik Das (Xavier Institute of Social Service, India), S. N. Singh (Xavier Institute of Social Service, India), Mahua Banerjee (Xavier Institute of Social Service, India), Shishir Mayank (Xavier Institute of Social Service, India) and T. Venkata Shashank (Xavier Institute of Social Service, India)
Copyright: © 2018 |Pages: 25
DOI: 10.4018/978-1-5225-5775-3.ch001

Abstract

Image data has portrayed immense potential as a resourceful foundation of information in current context for numerous applications including biomedicine, military, commerce, education, and web image classification and searching. The scenario has kindled the requirement for efficient content-based image identification from the archived image databases in varied industrial and educational sectors. Feature extraction has acted as the backbone to govern the success rate of content-based information identification with image data. The chapter has presented two different techniques of feature extraction from images based on image binarization and morphological operators. The multi-technique extraction with radically reduced feature size was imperative to explore the rich set of feature content in an image. The objective of this work has been to create a fusion framework for image recognition by means of late fusion with data standardization. The work has implemented a hybrid framework for query classification as a precursor for image retrieval which has been so far the first of its kind.
Chapter Preview
Top

Introduction

Information identification has been guided by a fresh perspective with the intensification of digital era. The popularity and growth of image capturing devices have prioritized images as a familiar media of communication for the mass (Korytkowski, Rutkowski, & Scherer, 2016). The image datasets have been archived and maintained as rich sources of valuable information (Ahmadian, & Mostafa, 2003). Traditional means of information identification with images have been based on text-based recognition system. The process used to map images with text-based annotations. However, text or keywords based mapping of images has insufficient information about image contents and is based on the perception of the individual performing the job of annotation (Walia, Goyal, & Brar, 2014). These are major drawbacks for recognition of information with images and to escalate the success rate.

The discovery of a object of interest or locating the region of interest in a picture or an arrangement of pictures, which has applications in confront acknowledgment and in addition to video conferencing frameworks, is a testing assignment and has been considered by numerous researchers (Das, & Walia, 2017). Once the test picture is removed from the scene, its gray level and size are standardized before putting away or testing. In a few applications, for example, distinguishing proof of international ID pictures or drivers' licenses, conditions of picture obtaining are typically so controlled that a portion of the pre-processing stages may not be essential.

Complete Chapter List

Search this Book:
Reset