基于多分辨率统计能级的SAR图像点目标检测

针对合成孔径雷达(syntheticapertureradar,SAR)图像点目标检测的困难,基于各层小波系数分布的统计特征,提出了一种在多分辨率统计能级上区分目标与杂波背景的方法。在非正交小波变换的基础上,定义各点的层间随机过程,进行各分辨率下的信息相关后,通过能量函数构造能量图像,并在能量图像上自适应地搜索合适的目标尺度窗口实现检测。实验结果表明,该方法适用于不同的杂波背景,能有效检测潜在的点目标,并在一定程度上保持了目标的形状。

Target detection of SAR images based on multi-resolution statistic energy level

According to the feature of point targets in SAR image, and based on the statistic characteristic of the wavelet coefficients at different scales, a method to distinguish between the targets and clutter background on multi-resolution statistic energy level is provided. Firstly, the raw data are decomposed by means of the nonorthogonal wavelet transform, then the random processes at each pixel position among different scales are defined, and the information at each pixel position among different scales is correlated. The energy image is built by the energy function, and point targets can be detected through the adaptive scale windows in the energy image. The results of experiments show that the method can apply to different clutter background and detect targets efficiently keeping the figures by much.