Moving vehicle detection in satellite videos by improved ViBe algorithm
Dudu Guo, Jie Zhang, Wen Lou, Liang Zhao · Remote Sensing Letters · 2023
The development of video satellite technology has provided new and reliable data for detecting moving objects. This study proposes a method for detecting moving vehicles in satellite videos based on an improved ViBe algorithm. Based on the spatial distribution characteristics of moving vehicles, the detection area in the video was divided into two parts. Next, the DeeplabV3+ semantic segmentation model was used to automatically extract the road area to remove the interference of various influencing factors outside the road. The detection rate of the ViBe object-detection algorithm for moving regions is improved by setting an adaptive dynamic foreground discrimination threshold. The proposed scale-centre distance constraint method was then used to suppress the interference of false moving targets on the road and solve the problem of large statistical errors caused by the stickiness of the adjacent vehicle targets. Furthermore, the validity of the proposed method is verified using the satellite image data acquired by ‘Jilin-1’ and ‘Skysat-1’.