Separation of Printed and Non-Printed sections using Modified Global Threshold approach from Medical Prescription Document Image

Abstract

Global threshold method is proposed in this paper, for separation of printed and non-printed sections from real time medical prescription images. There is no ground truth database available in the literature, for testing the performance of the proposed method.Hence, we have collected real time medical prescriptions from various Dentists, Gynecologists, Orthopedic Doctors, Cardiologists, Physiotherapists,Cancer Specialists, ENT Specialists, Urologists and Chest physicians. For experimental purpose, we have considered a total of 80 distinct sample medical prescription images.Then global threshold method has been performed on sample images for separation of printed and non-printed sections using different threshold values such as 0.15, 0.35, 0.55, 0.85 and 0.95. The proposed method obtained the segmentation accuracy of 46%, 38%, 38%, 38% and 36% respectively.Clearly visible sample images were obtained when the threshold value was 0.85. Hence, we have fixed 0.85 as the optimum threshold values for the experimental setup.