The Effect of Disturbance in UML Sequence Diagram Images on Image Plagiarism Detection Ability Using the Perceptual Hash Algorithm

Abstract

The Perceptual Hash algorithm has been widely used, especially to detect similarity or plagiarism of an image. Of course, this ability is applicable in education fields, such as plagiarism detection in collecting assignments for designing UML sequence diagrams. However, the algorithm's ability to detect image similarity may be affected by disturbances or changes in the image. Therefore in this paper, we researched the effect of disturbances in UML sequence diagram images on the ability to detect image similarity that the Perceptual Hash algorithm can do. The image disturbances we studied are rotating and skewing, which may be reasonable for students. These disturbances were tested on digitally generated images. The test results show that the Perceptual Hash algorithm's ability to detect the similarity of UML sequence diagram images is not affected by disturbances that occur in the image. The algorithm can still detect plagiarism attempts on UML sequence diagram images that have disrupted with a similarity detection rate above 60%.