摘要
中低分辨率SAR图像中河流、桥梁目标常存在断裂现象,且桥梁边缘线性特征不明显,现有桥梁目标识别方法难以解决上述问题。为提高中低分辨率SAR图像桥梁识别稳健性,提出桥梁在图像中呈现断裂、河流灰度一致性不高情况下的桥梁目标识别算法。对水域分割结果进行河流粗提取,将河流轮廓外扩部分的相交区域作为桥梁兴趣区,在保证兴趣区完整性的同时减少无效兴趣区;采用目标分割与Radon变换结合的方法检测断裂的桥梁主体;对候选桥梁附近的河流轮廓进行形状验证,消除山体等阴影造成的桥梁虚警。实验结果表明,该方法能够在复杂场景中有效地识别出多个水上桥梁目标,桥梁虚警率大大降低。
Recognition problems,which are caused by the rupture characteristics of river and bridges in medium-low resolution SAR images,cannot be solved by means of traditional algorithms.To improve bridge recognition robustness,we propose an improved algorithm which is adaptive to the bridge rupture and river＇s bad gray condition.The river contour is expanded,based on river coarse detection from segmented image,to obtain the intersectant area for the definition of region of interest（ROI）.This method can reduce invalid ROI with keeping the valid regions.Broken bridge body is detected by target segmentation and Radon transform.Bridge false alarm rate owing to mountain shadow etc,can be eliminated by rectifying the river contour nearby bridge.Experimental results show that the proposed method has good features in complex scene,reducing bridge false alarm rate at the same time.
出处
《雷达科学与技术》
2010年第6期510-515,共6页
Radar Science and Technology
关键词
合成孔径雷达(SAR)
中低分辨率
水上桥梁
兴趣区
虚警率
synthetic aperture radar（SAR）
medium-low resolution
bridge over water
region of interest
false alarm rate
