Abstract

In an educational environment, plagiarism is a crucial task that needs to be identified, in recent years all known journals and conferences, as well as universities, request a plagiarism report from students and researchers to prove the originality of published text or scientific paper. Plagiarism detection usually checks the text content via many of the platforms which are available for productive use reliably identifying copied text or near-copies of text and these systems usually fail to detect the images, and Files plagiarism since it is originally built for text mainly. In this paper, we suggest an adaptive, scalable, and extensible, robust method for image plagiarism which is tested in designs collect from department of architecture University of Technology, this method mainly compare the data (designs images) entered to the system with data sets saved in the database mainly these designs are saved as feature which is one of the artificial intelligence algorithms and match by using k-mean clustering and the similarity check is done with threshold used 40% which can be changed to an accepted levels when needed. Using the k-mean algorithm in clustering, which is a robust artificial intelligence clustering algorithm giving us a strong system that is not discarding any feature extracted from the image. In this paper, data sets consist of 45 samples as training images saved and used in the system as the system database and using 48 samples as testing images which consist of original and forgery designs. These testing images were evaluated with 100% matching rate and 81% matching accuracy rating.

Description