Autoregressive (AR) spectrum has the advantage of high frequency resolution over the Fourier spectrum in sea clutter analysis. In recent years, the multi-scale Hurst exponents are widely used in describing the AR spectrum of sea clutter due to their abundant clues about the local roughness of the sea clutter in different scale intervals. In this paper, a method based on whole-scale Hurst exponent of AR spectrum is proposed for weak target detection in the sea clutter background. The measured X- and S-band data sets are used to analyze the multi-scale Hurst exponent of AR spectrum and the results show that the difference degree based on the multi-scale Hurst exponent between sea clutter and targets varies with the scale intervals and the data band types. Then, the whole-scale Hurst exponent maximizes the difference degree by considering each scale of the multi-scale Hurst in the different data sets and thus for the convenience of weak target detection. Compared to the existing fractal methods and the traditional CFAR method, the proposed target detection method obtains a better detection performance in low SCR condition.

在海杂波分析中，自回归（AR）频谱比傅立叶频谱具有高频分辨率的优势。近年来，多尺度的赫斯特指数被广泛用于描述海杂波的AR谱，因为它们具有不同尺度间隔内海杂波的局部粗糙度的丰富线索。提出了一种基于AR谱全尺度Hurst指数的海杂波背景弱目标检测方法。利用实测的X波段和S波段数据集分析了AR谱的多尺度赫斯特指数，结果表明，基于海浪杂波与目标的多尺度赫斯特指数的差异程度随尺度间隔和数据段类型。然后，通过考虑不同数据集中多尺度Hurst的各个尺度，整个尺度的Hurst指数将差异程度最大化，从而方便了弱目标检测。与现有的分形方法和传统的CFAR方法相比，该目标检测方法在低SCR条件下具有更好的检测性能。