Currently, marine ship targets are often identified directly through sensing targets. As a result, the potential value of massive spatiotemporal data has not been fully utilized. Because relevant maritime authorities have new requirements for understanding the situation of naval traffic flows and target identity behavior, obtaining necessary information about maritime targets at the data level and further understanding the ship movement patterns at the characteristic level are the focus of the current study. From the perspective of system design, this study proposes a method for obtaining a number of feature sets by mining the ship trajectory data and then identifying unknown targets. In the process of constructing the multi-characteristic sets, this study focuses on the importance of the distribution characteristics of ship channels for target identification. This study proposes an improved similarity measure—a trajectory rasterization model method. Experimental verification based on the measured data set shows that the system design can successfully fulfill the requirements for the identification of unknown targets and has a certain universality.