Real-time People Counting System Using Low-resolution ToF Sensor based Blob Tracking Algorithm

This study proposes a human detection and tracking system using an 8×8 time-of-flight (ToF) sensor array for monitoring entry and exit in confined indoor spaces. Real-time depth data from the sensors is preprocessed through dynamic noise filtering using a reference height from the central region. A blob detection algorithm then isolates human targets, assigning each blob a unique identifier for multi-frame tracking based on its center coordinates stored in replay memory to capture movement patterns. Variations along the y-axis of the blob’s center are analyzed to determine entry and exit with high accuracy. The proposed method attains an average accuracy of 98.25% across diverse scenarios, such as single and multiple consecutive entries, path crossings, and entries involving additional objects.