As existing edge extracting method has its drawbacks in complex industrial environments, a laser edge image inpainting method based on pixel adjacency analysis is presented. The first step of the method is to obtain a label image by distance conversion and connected component mark for an edge image. In the label image, the background pixels whose minimum edge distances are less than a certain value are marked as edge candidate points while the rest background pixels are marked as independent connected regions. The second step is to re-label the edge candidate points based on the adjacency of real edge to connect the broken edges. And the final step is to categorize the noises as per adjacency and remove them respectively to inpaint the laser edge image. The experimental results indicate that this method can effectively inpaint a 8-pixel edge gap and remove larger noises; the introduced root-mean-square(RMS）error of centering measurement is 0.05 pixel, while the peak value(PV） error is 0.086 pixel, which show a steadily low value. Moreover, it takes less than 130 ms for an image to be inpainted one time in real-time. This method is applicable to industrial online centering measurement.