Drawing similarity measurement using object alignment

• Main Authors: Barfeh, Davood Pour Yousefian, Bustillos, Edward
• Format: text
• Published: Animo Repository 2019
• Subjects: Image processing—Digital techniques; Computer drawing; Shapes; Computer Sciences
• Online Access: https://animorepository.dlsu.edu.ph/faculty_research/10819; https://animorepository.dlsu.edu.ph/context/faculty_research/article/9102/viewcontent/Scholarly_Output___Hiring_of_Dr._Davood_Pour_Yousefan_Barfeh.pdf
• Institution: De La Salle University

Shape similarity plays important role in human perception. Humans always look for similarities between the shape of what they see and the shape of what they saw via comparison. Shape similarity has many applications in different fields such as cognitive science, medicine and computer science. This is a study in which a system is designed and implemented to measure similarities of two circular drawing shapes. The system follows different steps and techniques related to digital image processing such as; binarization, morphological operations and image segmentation. The novelty of the system comes from image alignment in which some transformation is performed to eliminate size dependency in similarity measurement. The transformation works on changing the origin point followed by stretching. In this study 30 pairs of circular shapes are considered. The results shows the system has 91.38% accuracy in average while maximum accuracy among 30 pairs of samples is 99.98% and minimum accuracy is 69.15%.