Automatic recognition of bridge above water in aerial images based on regional feature

Addressing the automatic recognition problem of a bridge above water in aerial images, an automatic recognition algorithm for a bridge above water in aerial image based on regional features is proposed. Firstly, aerial image binarization is executed based on the gradient mean square variance. The image noises are then removed based on pixel density. Next, the connective regions are labeled according to the pixels in binary image with a six neighborhood connection rule to obtain the water regional feature. Finally, the bridge is extracted. Experiments show this algorithm is effective for automatic recognition of a bridge above water in low-contrast aerial images taken at low altitudes.