First Arrival Picking of Exploration Seismic Data Using Pseudo-coloring and Color Image Segmentation

In this paper, a new method for automatic first arrival picking is introduced based on pseudo-coloring and color image segmentation of refracted seismic shot-gathers. The two-step method uses the spiral color transformation (known as the warm color mapping) to pseudo-color the shot-gathers. This is followed by a two-class k-means clustering algorithm to segment the given colored vectors representing the first arrival amplitudes and their timing. The proposed algorithm was tested on two challenging real seismic shot-gathers. With an assessment error of a 20 ms window with manual picking of such data, the proposed method outperforms the automatic method of Coppens’. Similarly, it also outperforms the semi-automatic method based on color segmentation of energy-ratios using Projections Onto Convex Sets (POCS). In both cases, the suggested method outperforms them by an overall average of 27.58%. Therefore, it is believed that the proposed method provides an accurate and a very good alternative to existing automatic first-arrival picking methods.