A proposed Figure Plagiarism Detection Techniques

Abstract

Plagiarism detection and prevention became one of the research and educational challenges. Many students and researchers tend to copy others work and ideas when doing research, projects, and assigned tasks. Most of the researches and papers that are meant for plagiarism detection are focusing on the plagiarism from text point of view. So far there is no significant works in figure plagiarism detection. One of these researches evaluate the effectiveness of two methodologies that capture the useful features in order to achieve higher classification accuracy from figure images, This work presents a novel method of figure plagiarism detection based on combination of two feature extraction methods : Discrete Wavelet Transform (DWT) and Discrete Cosine Transform (DCT) compression techniques .These methods are used to extract the essential features to build the Perceptual Hash which is used as a feature in figure images plagiarism detection. The combined model had an accuracy of 99.32% when the number of features is 32. Also another model that extracts the essential feature using the Discrete Wavelet transform (DWT), Discrete Cosine Transform (DCT) and Optical Character Recognition (OCR) is constructed. This model combines the three extractor features and in the classification stage uses Support Vector Machine (SVM) technique in order to achieve higher classification accuracy. The experimental results showed that the average classification accuracy is 84% for 32 features when the figure images caption and text on the figure were used, and 75% for 32 features when text on the figure without the figure images caption were used. Farther investigations were recommended using large corpuses and different classification. This study recommends the inclusion of the text in the figure and text of caption in detecting figure plagiarism.