Local Angle Extraction and Noise Attenuation for Seismic Image Using Contourlet Transform

We propose contourlet-based algorithms to extract local angle from seismic migrated images and attenuate seismic noise. Steerable pyramid and projection-filter-based complex wavelet transform have been used to extract seismic local angle, while curvelet transform and complex wavelet transform have been used to attenuate seismic noise. However, previously mentioned transforms either have a fixed number of directions, making the local angle extraction rough and not flexible enough for seismic noise attenuation, or are significantly overcomplete, leading to poor practicability for use in super-tremendous seismic data processing because of computational expense. Contourlet transform is a unique transform with flexible number of directions at each scale while achieving nearly critical sampling. Thus, we take advantage of the contourlet transform to extract local angle and attenuate seismic noise. Experiments show that the proposed algorithms can extract accurate local angle computationally efficient, and provide superior noise attenuation results with minimal impact on the desirable signal, which is illustrated using a stacked-data example.