地震资料的高阶伪希尔伯特变换边缘检测

Seismic data edge detection based on higher-order pseudo Hilbert transform

构造了一种用于边缘检测的高阶伪希尔伯特变换,它吸收了时频分析的局部化特性和传统希尔伯特变换的思想,调节其时窗宽度和阶数可以实现不同的输出.信号模型及实际图像的处理结果表明,高阶伪希尔伯特变换能提取出各种类型的边缘信息,边缘表达完整、定位准确.利用这一方法对实际地震资料进行了处理,它能突出地震资料的奇异信息.可用于描绘地层精细结构,地层的识别和追踪,裂缝、河道和断层的检测.

A high-order pseudo Flilbert transform （FIOPFIT） which can be applied to edge detection is presented in the paper. HOPHT combines the localization property of time-frequency analysis and the idea of traditional Hilbert transform （HT）. Its output depends on adjustment of the width of window function and the order. The simulation of the synthetic signal model and real image processing demonstrate that HOPHT can extract information of various ed- ges. The extracted edges are accurate and integrated. HOPHT is applied to real seismic data processing. The results show that it can highlight singular information of seismic data. It can be applied to delineating the fine structure of strata, the layer recognition and detecting fractures, channels and faults.