An approach for keyframe extraction using AdaBoost is proposed which is based on foreground detection. The aim of this approach is to extract keyframes from sequences of specific vehicle images of lane vehicle surveillance video. This method utilizes integral channel features and the area feature as the image feature descriptor, combined with training an AdaBoost classifier. The experimental results on real-road test video show that the algorithm presented in this paper effectively selects the most distinct and clearest image for a sequence of vehicle images which begins counting when a motional vehicle enters into the surveillance area and ends when it leaves. Compared with other methods, it has increased the effectiveness and precision for keyframe extraction of lane vehicle surveillance video and achieves more effective compression of video analytical data for lane vehicle surveillance.