A Preliminary Study on Extracting the Collarette Features from the Iris Images

• Published in: 2022 13th Asian Control Conference (ASCC) pp. 1541 - 1543
• Main Authors: Park, Miso S., Park, Sungjoon, Hur, Wang-Jung, Park, Seong-Il, Park, Hyun-Jung, Yoo, Ho-Ryong
• Format: Conference Proceeding
• Language: English
• Published: ACA 04-05-2022
• Subjects: boundary estimation; collarette; computer vision; Feature extraction; Hospitals; Image edge detection; Iris; Manuals; Programming; Transforms
• ISSN: 2770-8373
• DOI: 10.23919/ASCC56756.2022.9828348

Collarette, a round or zigzag-shaped line that separates the iris's central pupillary zone from its peripheral ciliary zone, is a key feature in iridology. In this study, we used computer programming to determine the exact length and area of the collarette. We used iris images from 204 patients who visited the Clinical Neuroscience Center at Daejeon Korean Medicine Hospital. We computed the edge map and used the Hough circular transform to fit circles to the iris and pupil edges to extract the iris and pupil boundaries. The pupil boundary could be determined after determining the iris boundary by focusing on the interior of the iris. The collarette boundary was estimated using Python programming. Because the collarette boundary was difficult to extract automatically, we also used manual labeling. The average ratio of the collarette to the length from the pupil border to the outer border was calculated to be about 0.27, which is close to the commonly accepted value of 0.3. Meanwhile, the average collarette area to total iris area ratio was around 0.15. The collarette boundary's average radius has a correlation coefficient of 0.70, a root-mean-square error of 7.1 pixels with the hand-labeled boundary. More research is required to improve the accuracy of the algorithm.