I have developed a new automatic picking algorithm that can be applied to many picking problems in seismic data processing. This algorithm does the initial picking by solving a constrained non-linear optimization problem with a fast search algorithm. Then, by linearizing the model of data, the objective function defined in the optimization process is approximately reduced to a quadratic form. The residual corrections are obtained by solving a linear equation.
I applied this algorithm to several practical problems, including dip-picking, event-picking, well-log interpolation and velocity picking. Examples with field data show that the results are reasonably accurate and reliable.