| Radon transform is employed commonly in seismic data processing and interpretation, especially in the events identification and estimation, multiples attenuation, missing seismic traces reconstruction, velocity analysis and wave field separation. The least squares inversion scheme, being low in the transform resolution and computation efficiency, is the most ordinary method to implement the Radon transforms for linear, parabolic and hyperbolic type. Therefore, a lot of works need to be studied in enhancing the transform resolution and the computation efficiency.The procedure of Radon transform realized by the least squares inverse method is analyzed from the definition of dip overlapping, the question about what is Radon transfonn resolution and how to resolve it are elucidated. The damping least squares normalized method might lead to low resolution because of normal matrix being constant. High resolution linear and parabolic Radon transform is derived by modifying normal matrix—giving the sparse solution of least squares Radon transform. Hyperbolic Radon transform which is not time-invariable should be done in time domain, while the solution and restriction of high resolution algorithm is obtained in T-q domain.The programs of linear Radon transform algorithm are realized in the time domain and frequency domain respectively, and models are designed to balance the two algorithms, the calculation results testify that the high resolution algorithm in frequency is better. Least squares algorithm and high resolution algorithm about parabolic Radon transform are devised in the frequency domain; the calculation of model data proves the high resolution is more effective. The program of hyperbolic Radon transform is realized in the time domain, the result testifies hyperbolic Radon transform can do very well in separating pressure wave and shear wave. The processing of real data shown in this paper proves these methodsabove-mentioned are effective and practical.
|
PDF Full Text Request