基于区域特征的航拍图像水上桥梁自动识别  

Automatic Recognition of Bridge Above Water in Aerial Images Based on Regional Feature

作　　者：董银文[1] 苑秉成[2] 石钊铭[1] 

机构地区：[1]海军工程大学电子工程学院,武汉市解放大道717号430033 [2]海军工程大学兵器工程系,武汉市解放大道717号430033

出　　处：《武汉大学学报（信息科学版）》2013年第9期1073-1077,共5页Geomatics and Information Science of Wuhan University

基　　金：湖北省自然科学基金资助项目(2011CDC023)

摘　　要：针对航拍图像中水上桥梁目标的自动识别问题,提出了基于区域特征的水上桥梁自动识别算法。首先对航拍图像进行基于梯度均方差的图像二值化;再进行基于像素密度的二值图像去噪;然后进行基于像素的二值图像连通区域标记,区域标记算法采用6邻域连通规则进行标识,能够有效获取水域的区域特征;最后进行桥梁的精确提取。实验结果表明,该算法能够有效地识别低对比度、低空侧拍等复杂航拍图像中的水上桥梁目标。Addressing the automatic recognition problem of a bridge above water in aerial im- ages, 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 gra- dient 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 ex- tracted. Experiments show this algorithm is effective for automatic recognition of a bridge a- bove water in low-contrast aerial images taken at low altitudes.

关 键 词：航拍图像 桥梁识别 区域分割 区域标记 梯度均方差 

分 类 号：P237.3[天文地球—摄影测量与遥感]