Vehicle identification is widely used in route planning, safety supervision and military reconnaissance. It is one of theresearch hotspots of space-based remote sensing applications. Traditional HOG, Gabor features and Hough transform andother manual design features are not suitable for modern city satellite data analysis. With the rapid development of CNN,object detection has made remarkable progress in accuracy and speed. However, in satellite map analysis, many targets areusually small and dense, which results in the accuracy of target detection often being half or even lower than the big target.Small targets have lower resolution, blurred images, and very rare information. After multi-layer convolution, it is difficultto extract effective information. In the satellite map data set we produced, the target vehicles are not only small but alsovery dense, and it is impossible to achieve high detection accuracy when using YOLO for training directly. In order tosolve this problem, we propose a multi-feature fusion target detection method, which combines satellite image andelectronic image to achieve the fusion of target vehicle and surrounding semantic information. We conducted a comparativeexperiment to demonstrate the applicability of multi-feature fusion methods in different detection models such as YOLOand R-CNN. By comparing with the traditional target detection model, the results show that the proposed method hashigher detection accuracy.
展开▼
