Human detection and tracking in low-resolution infrared images for smart home applications

This study aims to develop a robust method for detecting and tracking individuals in indoor environments using low-resolution infrared (IR) array sensors, contributing to smart home systems for energy management, security, and user comfort. Traditional tracking methods, such as Kalman filters, struggle with noisy 32x32 IR images due to low image quality. The proposed method addresses this limitation by using displacement measurement between bounding boxes across frames to assign consistent IDs to individuals. Additionally, image quality is enhanced using median filtering, contrast stretching, and multi-level thresholding to handle overlapping individuals. The experimental results demonstrate that the proposed method effectively manages occlusion scenarios and noise in infrared data, outperforming traditional methods in terms of accuracy and reliability. The proposed method provides a practical solution for individual detection and tracking in low-resolution IR images, making it suitable for real-world smart home applications. This method is beneficial for smart home systems, improving energy management, security, and user comfort through accurate individual detection and tracking.