This research purposes a digital image processing tool for cephalometric landmark localization (phase 1). This type of landmarks is essential in that orthodontists use calculate them to distances, degrees or length ratios for orthodontic treatment. This research is interested in automatically localization or 17 Cephalometric landmarks, 40 in total. These landmarks are used as a preliminary analysis and calculation of orthodontist at orthodontics department of Chulalongkorn University. The landmark detection process starts with normalization of radiograph images. Many difference image processing are applied for individual landmark. Our experiments of automatically detecting 17 landmarks per image, from 23 images, indicate very high positive results. Based on orthodontists’ acceptable range, that is 5 millimeters from ground truth locations, 76.21% are correctly detected while 67.52% are within 3 millimeters. False detection rate is 23.79%.
