Title Development of an Algorithm for Detecting Pedestrians and Vehicles in Nighttime CCTV Footage of an Apartment Complex's Outdoor Space
Authors 오경철(Oh, Kyoungcheol) ; 김창민(Kim, Changmin)
DOI https://doi.org/10.5659/JAIK.2025.41.11.237
Page pp.237-247
ISSN 2733-6247
Keywords Nighttime surveillance video; CCTV; Moving object detection; Image processing
Abstract Unnecessary lighting in apartment complex outdoor spaces leads to energy waste and light pollution, causing dissatisfaction among residents. By using CCTV to detect moving pedestrians and vehicles, lighting can be activated only when necessary, improving energy efficiency without replacing existing systems or infrastructures. This study suggests a method for detecting moving pedestrians and vehicles using nighttime CCTV from apartment complex outdoor spaces. The proposed method mixes frame differencing with an improved background subtraction algorithm, setting its sensitivity based on lighting conditions. Issues such as low lighting, shadows, and noise, which are common at night, are overcome by this. Also, Video preprocessing techniques such as resolution compression, contrast enhancement, and noise filtering are applied to improve detection performance. This algorithm is validated using real CCTV collected from two residential complexes in Korea. The proposed method yields up to a 26% improvement in precision compared to existing methods. This research shows the detection performance while protecting personal privacy by avoiding data-intensive training processes. The results mean that this approach enables energy savings in residential outdoor lighting systems.