摘要
图像分割和目标方位角估计是进行SAR (SyntheticApertureRadar)图像自动目标识别的重要步骤。文章提出了一种基于MRF (MarkovRandomfield)模型的SAR图像分割算法 ,利用ICM (IterativeConditionalMode)局部优化方法 ,获得MAP (maximumaposteriori)准则下的图像分割结果 ,将图像分割为目标、阴影、背景三部分。然后确定目标离雷达最近的点 ,从而得到目标的主导边界 ,并估计出目标的方位角。用MSTAR (MovingandStationaryTargetAcquisitionandRecognition)数据进行实验 ,估计方位角的准确性与现有算法的结果相比 。
Image segmentation and target aspect estimation are very useful for automatic target recognition in SAR imagery Based on the MRF model, we first segment the image using ICM algorithm According to the MAP criteria, the image is segmented into target, shadow and background Then the target pixel closest to the sensor is obtained, and we can get the primary edge and estimate the target aspect angle based on it These algorithms are applied to SAR imagery from the MSTAR datasets, and the result is better than by other algorithms
出处
《国防科技大学学报》
EI
CAS
CSCD
北大核心
2001年第5期74-78,共5页
Journal of National University of Defense Technology
