| To improve the processing efficiency of video data used in intersection operation evaluation and to realize real-time diagnosis of intersection problems, a structured video-data based method is proposed, with headway as the primary analysis object. First, an algorithm based on the reproducibility of the flow curve is presented to search for the peak period. Then, the headway curve is used to realize the synchronous comparison of video data and signal control schemes. A comprehensive evaluation method of intersection efficiency is proposed from two aspects of signal control scheme and lane. Finally, the dynamic time warping and the grey clustering method are utilized to identify five types of lane headway curves based on actual data, with the corresponding problems and optimization schemes listed. The method provides a solution for problem diagnosing of inefficient lane and phase.