Clustering analysis is a classical approach to mine the useful information from huge maritime database. In order to do clustering for large amounts of samples, considering most adjustments of clusters happen in the initial stage of iteration in the clustering process of k-means algorithm, this paper proposes an accelerated clustering analysis algorithm based on cluster’s threshold-adjusted. Experiments proved that the improved algorithm can not only obtain good clustering result, but also accelerate the speed of clustering. Finally an application of clustering is given for designing routes from the marine database，which is very important for maritime intelligent transport system.