Abstract
The common offset gathers extracted from common image gathers can be seen as multiple observations about the same underground image. Due to migration errors, the real image positions deviate from the true positions with lateral and vertical misalignments. This paper proposes a seismic image warping technique to enhance the common image gathers. This method aligns different common offset gathers point to point in both the lateral and vertical direction. In this paper the 2D local normalized cross-correlation can be used to measure the similarity of the input gather and the model gather at the same temporal and spatial position. By assuming the 2D cross-correlation computed at various lateral and vertical relative lags satisfies a continuous convex function, the derivation method can be used to estimate the non-integer correction values corresponding to the maximum correlation at the given position. After that the accurate image amplitude can be estimated by the bilinear interpolation. Compared with the traditional gather flattening method, the proposed method can estimate the lateral and vertical correction values of the common offset gathers and obtain the accurate imaging amplitude in the sub-pixel domain. The processing results of the 2D synthetic and real dataset demonstrate the proposed method can enhance the event consistency of common image gathers and improve the resolution of the stack.