The Technology Of Recognition Of Bridges Over River In Aerial Image

Automatic target recognition has important application value in the field of military and civilian.Bridge, as a typical representative of the man-made object, has brought many scholars’ concern.According to the aerial image, we proposed a knowledge-based target bridge detection method. That is using bridge prior knowledge to guide the each processing step of the bridge recognition. Because of the different aerial height, make the different bridge feature. So the bridge recognition in aerial image is divided into two parts: bridge recognition in high altitude aerial image and bridge recognition in low altitude aerial image.In bridge recognition of high altitude aerial image, bridge occupies a small percent in the high altitude aerial image, and is form of a long rectangular. So we should use river to help to recognize bridge. First, using the method of image gradient and threshold segmentation to segment aerial image according to the feature of river’s gradient. Methods through the combination of area method and gray feature phase to remove the interference region in segmentation image. Then, using the region growing algorithm repair the river region. Finally, determine the bridge position according to the fracture location of river.Bridge occupies more percent in low altitude aerial image than in high altitude aerial image. Using the character of river and bridge structureidentify the bridge. Segment the low altitude aerial image to remove some interference region with the improved Otsu segmentation algorithm. Detect the contour line bridge with the straight line detection method based on Hough transform and Freeman chain code method. Finally, according to the feature of river and bridge, using the texture feature of river and geometrical feature of river recognize the bridge.According to the above bridge recognition algorithm, developed the aerial image recognition of bridge system. This system realize the storage of aerial image data, bridge recognition in aerial image and the storage of the result of recognition. Design the experiment data of aerial image for the bridge recognition experiment.