摘要： 高分辨率SAR图像目标识别系统首先要检测出图像中可能的目标区域。该文利用累积概率分布及K-S距离方法确定海洋SAR图像中可能的目标区域，并提出了综合分析整幅图像及目标区域图像确定阈值的方法，利用此方法把成群像素区域二值化为二值图像，二值化结果优于常规方法,特别适用于弱对比度的SAR图像目标检测识别。

关键词: 合成孔径雷达, 累积概率分布, 目标检测

Abstract: Object recognition of high resolution SAR image is useful to the commander, the first step of the task is to find the interesting area in the image. The paper uses accumulative probability distribution and K-S distance to detect objections in sea SAR images for the first time. It also gives a method to find the threshold by analyzing both original image and object areas. With this method, it can get a more useful binary image than the ordinary methods. It's especial useful in the less sharp contrast SAR image object detection and recognition.

中图分类号: 

• TN957.52