Abstract: In image-based object detection, the airport is an important kind of object, and automatic recognition of it is of great significance. Aimed at the difficulty of general detection algorithms to correctly extract edge information from complex aerial images, a new object detection algorithm based on style transfer is proposed to enhanced edge feature extraction. The image noise was suppressed by using the generative adversarial network. The noise-suppressed image was transported to an edge detection algorithm to highlight the edge features. The highlighted images were used to complete the airport location detection through the object detection algorithm. In the airport object detection experiment, the object detection algorithm combined with the edge feature extraction approach proposed in this paper has higher precision than the original object detection algorithm. The average precision of YOLOv5 and the proposed feature extraction fusion algorithm achieves 97.7%. Experimental results demonstrate that this feature extraction approach has a good effect on airport object detection.