Anomaly detection is useful in diverse domains including fault detection system, health monitoring, intrusion detection, fraud detection, emotion recognition, cancer detection, animal rescue, detecting ecosystem disturbances, and event detection in sensor networks. Thermal image is a widely used night vision technology. Anomaly detection using thermal image features has been proposed in this work. Three major classes of features, namely textural features, color features, and shape features, have been extracted. Correlation model has been used for detecting anomalies. Thermal image of perishable objects has been analyzed, and the evaluation result confirms the hypothesis. It is found that using a set of features while using correlation as similarity measure the achieved average recall is 76.06%.
C. Mishra, Bagyammal T., and Parameswaran, L., “An Algorithm Design for Anomaly Detection in Thermal Images”, in Innovations in Electrical and Electronic Engineering, M. N. Favorskaya, Mekhilef, S., Pandey, R. Kumar, and Singh, N., Eds. Singapore: Springer Singapore, 2021.