Seismic Edge Detection Based On Windowed Hilbert Transform In Time Domain

Hilbert Transform is an important theory tools in analyzing and processing signals. Hilbert Transform is applied to process seismic data by Taner et al in 1979. It is extensively applied to calculate instantaneous characters. Because Hilbert Transform is sensitive to noise and weak in antinoise ability, Generalized Hilbert Transform, we also can call it windowed Hilbert Transform in frequency domain, has been proposed by Yi Luo in 2003. In the paper, we introduce a new generalized Hilbert Transform——windowed Hilbert Transform in time domain.We introduce the concept, the fundamental idea and the algorithm achievement of windowed Hilbert Transform in time domain. It is applied to process real seismic materials.1. Seismic complex trace analysis. We introduce several algorithms to calculate instantaneous frequency, and compare them. There are other instantaneous characters, such as, instantaneous bandwidth, instantaneous dominant frequency, and instantaneous quality factor etc. We bring in the calculation and the effect of these instantaneous characters. And they are applied in seismic data processing.2. Seismic edge detection. In the image especially in seismic data, the edge is a very important character. However, the materials' Signal Noise Ratio (SNR) seriously influences the result of seismic edge detection in real seismic data processing. So we must reduce noise before running edge-detection algorithms. In this paper, wc introduce a noise-reduction method, edge-preserving smoothing (EPS). It can preserve the abrupt change while successfully reducing noise.