基于SVM的SAR图像中水上桥梁目标的检测

针对SAR图像中桥梁目标的特性,提出了一种基于统计特性和先验知识相结合的桥梁目标检测方法。通过对SAR图像中桥梁和背景的分析,首先对河流区域进行特征提取,再利用支持向量机方法对数据进行训练建模,通过训练后的模型对SAR图像中的河流进行分割,最后在分类后的二值图中按方向累加能量最小准则进行桥梁目标检测。基于真实SAR图像的实验结果显示,此方法不需要对SAR图像进行复杂的预处理,有强的抗斑点噪声性,能快速检测SAR图像中的桥梁目标。

SVM-Based Detection of Bridge Object Over Water

Based on analyzing bridge and background feature in the SAR images, a new detection method combined statistical property with knowledge of bridge object is presented in this paper. For the high resolution synthetic aperture radar(SAR) images, the background of the target image is very complicated, and it is hard to detect the real targets from the false targets in some images. In addition, SAR images are perturbed with speckle noise that result in difficulties in the detection task of the bridge targets. So the feature of river domain is extracted firstly. Second, support vector machine(SVM)-used made model via training example data to segmenting river region in SAR image. The last based on direction energy function identify bridge target in the binary image of river class. Experimental results show that the proposed method has good features for detection bridge target.