Image Segmentation Using Clustering Method for Human Eye Localization

In this work, we developed a MATLAB program that implements image segmentation using clustering algorithms to accurately locate human eye positions. The implementation utilizes k-means clustering for partitioning image pixels based on color or intensity features, creating distinct regions for subsequent analysis. Following segmentation, we apply morphological operations such as opening and closing to remove noise artifacts and perform image smoothing using Gaussian filters to enhance result accuracy.

The eye localization process involves analyzing segmented regions through feature extraction techniques including region properties calculation (area, centroid, eccentricity) and contour analysis. We implemented threshold-based filtering to isolate candidate eye regions based on size and circularity metrics. The algorithm incorporates template matching and geometric constraints to verify eye pairs, ensuring robust detection under varying illumination conditions.

This computer vision research represents a critical step in image understanding, providing foundational work for subsequent analyses in face recognition systems and human-computer interaction applications. The modular code structure allows for parameter optimization and integration with deeper learning frameworks for enhanced performance.