Image segmentation is a critical and vital component in many application domains. The proposed work focuses on performing segmentation in thermal images using neutrosophic idea of truth, indeterminacy, and falsity membership. The computations of these memberships are then pre-processed using α-mean and β-enhancement to minimize the indeterminacy. Then, as a part of the convergence criterion, the gamma clustering is used to define segmented parts. Thermal imaging is a visual representation of different colour regions based on thermo gram. These images are not affected by variation in illumination is commonly used because it uses the intrinsic emissivity of thermal radiation from the human body. Drowsy driving application uses this significant advantage as both hardware and software components need to adapt both high intensity of daylight and zero intensity of night light situations. Drowsiness factors such as yawn and eye-blink are both considered for efficient performances outcomes. On experimental analysis of the proposed methodology provides an unbiased result with the drowsy detection accuracy of 97.01%.

Blink detection, Clustering, Drowsy driving application, Neutrosophic Sets, Thermal Images, Yawn detection