摘要
根据高分辨率SAR图像中桥梁目标的特点提出了一种新的桥梁识别算法,主要由水体分割、感兴趣区域确定和桥梁识别三个步骤实现。通过直方图均衡,方便了分割阈值的选取;通过数据采样,提高了识别的有效性;通过目标特征的抽取,保证了识别结果的准确性。识别过程在初级处理中,利用快速算法得到感兴趣区域;在后续处理中利用目标特征对该区域内的潜在目标进行识别,取得了良好的效果。
Base on the characteristics of bridge in high resolution SAR imagery, we present a new bridge recognition method, it is mainly composed of image segmenting, interesting area locating and bridge recognition. By the histogram equalization, we make it more convenient to determine the segmentation threshold; by the image data sampling, we enhance the efficiency; by the bridge characteristics extraction, we make the result more creditable. In primary processing, we get the interesting areas where bridge （s） are obtained by fast arithmetic. In the subsequent processing, guided with the bridge＇s character, we identify the bridge in the interesting areas. In the processing of automatic target recognition in high resolution SAR imagery, accuracy is needed on the one hand, and required is we needed on the other hand. So in this paper, the automatic bridge recognition algorithm focuses on not only the veracity, but also the rapidity. Veracity is ensured by the summarized characters of bridge in high resolution SAR imagery; rapidity is ensured by the methods of data sampling and false interesting areas wiping off. At last, in the recognition part, efficacious criterions are used to distinguish real target. Result shows the ability of this approach in high resolution SAR imagery.
出处
《遥感学报》
EI
CSCD
北大核心
2007年第2期177-184,共8页
NATIONAL REMOTE SENSING BULLETIN
基金
中国科学院电子学研究所种子基金项目(编号:2237-07)
关键词
SAR图像
图像分割
目标检测
桥梁识别
SAR imagery
image segmentation
target detection
bridge recognition
