Binocular Pupil Detection Based on Adaptive Gabor Filter

In modern medicine, clinical examination of the pupil is fundamental in ophthalmology. The results of these examinations are crucial for assessing the severity of certain diseases, guiding the development of new medical devices and medications, and determining the suitability of specific medical procedures. However, existing circular detection operators and threshold segmentation algorithms are affected by disturbances such as eyelashes and shadows, leading to poor detection accuracy. To solve this problem, this paper proposes a method for pupil diameter detection based on adaptive Gabor filtering. The method first utilizes unequal weight centroid method to roughly locate the center of the pupil. Subsequently, adaptive Gabor filter is applied to reduce noise and segment the pupil region. The image is then binarized, and a neighbor tracing algorithm is employed to search for the pupil contour. Finally, ellipse fitting is performed to obtain the pupil diameter parameter. The experimental results indicate that the algorithm effectively removes interference factors such as irises and eyelashes, while maximally preserving the required pupil information, thereby enhancing detection rate and accuracy. This method holds significant application value in filtering and noise reduction.