Moving object detection in satellite videos based on an improved ViBe algorithm

This paper studies moving object detection in satellite videos, which plays a significant role for large-scale video monitoring and dynamic analysis. Due to the tiny targets, complex background, and completely or partly occlusion, moving object detection accurately from each image frame is difficult and challenging. In order to tackle the issues, we propose an improved Visual Background Extractor algorithm (ViBe) with an improved Canny edge detection operator. First, the improved Canny operator based on iteration method is employed to find the optimal high and low thresholds and extract more edges information of satellite videos. Furthermore, the eight pixels are replaced with twenty-four pixels and the times of target threshold is introduced in ViBe detection to increase the selection range of backgrounds. The experiments are tested in SATSOT datasets, where the results verify that the proposed method is more robust in complex background with interference, turning and rapid motion.