将平滑度约束引入特征匹配是已知的可以实现超强鲁棒匹配。 然而，这样的匹配方案既复杂又缓慢，使得它们不适合于视频应用。 本文提出了GMS（基于网格的运动统计），一种简单的方法，将运动平滑度作为一个统计量，进行局部区域的匹配。GMS可以将高匹配数字转换成高匹配质量。 这提供了一个实时、超强的匹配系统。 评估低质量、模糊的视频和广泛基线显示，GMS始终如一地优于其他实时匹配器。(The introduction of smoothness constraints to feature matching is known to achieve super robust matching. However, such a matching scheme is both complex and slow, making them unsuitable for video applications. In this paper, GMS (motion statistics based on mesh) is proposed, which is a simple method to match the local area with smoothness of motion as a statistic. GMS can convert high matching numbers to high matching quality. This provides a real-time, ultra strong matching system. Assessing low quality, blurred video and extensive baseline display, GMS is consistently superior to other real-time matches.)