The key link of seismic signal processing is first break picking，whose accuracy and

efficiency constitute an important index of data processing．As in conventional seismic data，

especially in shallow seismic data，the signal-to-noise ratio is relatively high，the paper discusses the

improvement of the energy ratio method, i.e., the realization of the automation of the process under

the prerequisite of ensuring the accuracy．Based on the traditional energy ratio method，the authors

have studied two methods which use edge detection and stability factor to solve the "multiple first

arrive" phenomenon，and proposed a multiple-window energy ratio method to identify first arrival

time between refraction wave and direct wave. To solve the problem of accuracy and efficiency，the

authors suggest using the singular values of processing methods and variable phase shift technique．

The model and actual data processing results indicate that, with the integration of energy，window

and phase factors, the improved means of conventional seismic data has a better pick accuracy and

automation, and the computational efficiency is rapid too.