Objectives  The road network is the skeleton of a city and the road pattern refers to the shape of roads that can be clearly named. Identifying road patterns helps to understand the structures of cities and find more potential knowledge. However, urban road networks are complicated, and it is very difficult to identify the road structure.

  Methods  A link-graph-based method is proposed for recognizing the grid pattern in road networks. This method uses the link graph to represent road networks, in which nodes represent road segments and links represent intersections. It converts recognizing the grid pattern in road networks into searching for loops in the link graph. Five parameters are defined to select nodes and links that would match the grid conditions in the graph. The breadth-first search algorithm is used to search the loop in the link graph.

  Results  By mapping links and nodes in the link graph back to road segments, we can identify the grid pattern in road networks. This method can still work in more complicated road networks.

  Conclusions  The result shows that the proposed method can effectively recognize the grid pattern in road networks. Other road patterns can be further identified through analysis of the characteristics of the road connection.