To meet the requirements of a laser firing simulation system for high speed and high accurate location of laser spot centroids, a novel method based on the video sequential images is proposed for the spot detection and spot centroid location. The method firstly detects the firing events by the subtraction image between every two sequential frames and by the estimated noise parameters. Then, it uses the noise estimation to determine the segmenting threshold of the spot and combines morphologic filtering techniques to extract the spot region out from the background, meanwhile reducing the noises inside and outside of the window. Finally, one high resolution image is generated from 4 subtraction images to decrease the image noise and computing errors and to improve the locating accuracy of the spot centroid. Experimental results indicate that the spot centroid localization precision of laser spot and the average measurement stability of the proposed method are superior to that of the conventional method. The spot centroid localization precision has been reached to sub-pixel level, and the average measurement stability is 0.000 49, far better than the conventional 0.002 97. The method in this study is conductive to improving the performance of laser firing simulation systems.