介绍了基于高阶统计量（HOS）的一种识别地质孔洞的新方法，该方法通过对探地雷达信号进行双谱时频分析（BTFA），提取由孔洞引起的信号高斯性变化来识别地下溶洞。理论分析和实验结果表明，HOS方法能够有效地识别出高斯信号强干扰下的非高斯微弱信号，对孔洞识别的结果直观、清晰。

Based on the higher －order －statistics (HOS), a new method for identifying geologic cavities was introduced. By bispectrum time －frequency analysis (BTFA) to GPR signals, the variation of Gaussian properties generated from the geologic cavities to identify it was obtained. Theoretical and experimental results indicated that, the higher －order －statistics method is capable of recognizing the non－Gaussian signals under the disturbance of the strong Gaussian noise, and it can obtain clearer results than traditional methods when being applied to detect cavities.