Pupil detection based on constraint localization

Pupil detection is a critical issue in human-robot interaction applications in the field of robotics. A pupil detection method based on constrained positioning was proposed, aiming at the problem of blind search in detection algorithms. Firstly, the face detector was used to detect and locate the face area, and then the feature points of the eye area were found using the detector. Histogram equalization and normalization were performed to improve the contrast between the detected and the non-detected regions. Finally, filtering and edge detection were performed to determine the specific eye area, and then the Hough circle transform was used to detect the pupil. Experimental verification using self-made data sets shows that the method based on constrained positioning avoids global search, improves accuracy, and can detect pupils on faces in different poses.