Change Detection of Wide-Field-of-View Video Images in Foggy Weather Based on Graph Structure Similarity Metrics

Under foggy weather, Hawk-Eye surveillance has a wide field of view and low illumination of the surrounding environment. Its captured surveillance video image noise, visibility bottom, the monitoring system is difficult to detect the wide field of view under the small area of change. To address this issue, this paper proposes a change detection method based on graph structure similarity metrics. First, video frames from two different time points are pre-processed using a dehazing technique based on the color attenuation prior to enhance image quality. Then, based on the graph structure similarity metric, a similarity structure graph is constructed for each block and a K-neighborhood similarity structure graph (KNN) is constructed for the images before and after the moments respectively and the difference images are computed. Finally, the final change detection results are generated by clustering analysis. The experimental results on the real foggy scene dataset show that compared with the existing algorithms, the proposed method is able to effectively detect small regional changes in the video images before and after the foggy conditions, and exhibits strong robustness.