Intelligent systems to forgery image detection based on the edge characteristics using soft computing techniques

Divyashree, S and Narendra, V G and Kamath, Priya (2018) Intelligent systems to forgery image detection based on the edge characteristics using soft computing techniques. International Journal of Engineering and Technology(UAE), 7 (4). pp. 4493-44497. ISSN 2227-524X

[img] PDF
6001.pdf - Published Version
Restricted to Registered users only

Download (456kB) | Request a copy

Abstract

Detection of forgery in an image is an important aspect and very much essential in order to maintain the authenticity and the privacy of the information that could be interpreted from the image. In these days images could be manipulated using various tools and techniques, so its originality is lost, because these images may be used in medical diagnosis for monitoring the patients’ health care, in smartphones for user authentication, in forensic investigations where images serve as legal evidence. In such cases it is necessary to get the correct details from an image or else it may be misused. Hence there is a need to detect the altered images and prevent the intended people in extracting the false details from the manipulated images. This research focuses on a technique to detect the forged images (tampered images) based on the edge characteristics, that are supported by the functions such as Canny edge detection and Hough transforms, to extract a feature vector for the Gaussian filtering detection. Better performance is obtained by using Random Forest classification for K-Means clustering, SGD classifica-tion for Density based clustering and hence detect the images that are forged

Item Type: Article
Uncontrolled Keywords: Forgery Image Detection; Authenticity; Canny Edge Detection; Hough Transforms; Gaussian Filtering.
Subjects: Engineering > MIT Manipal > Computer Science and Engineering
Depositing User: MIT Library
Date Deposited: 30 Jan 2019 10:51
Last Modified: 30 Jan 2019 10:51
URI: http://eprints.manipal.edu/id/eprint/153085

Actions (login required)

View Item View Item