Abstract: Aimed at the problem that the QuickShift algorithm needs to manually specify the field value in the migration process, which leads to poor performance on complex datasets, an improved shared nearest neighbor density clustering algorithm (QS-SNN) is proposed. The proposed algorithm was based on shared nearest neighbors (SNN). It calculated the similarity of each pair of points in the dataset, obtained the local density matrix of the sample points through the similarity measurement. The sample points were quickly shifted in the SNN field, so that the final clustering result was obtained. Experiments on multiple datasets show that the QS-SNN algorithm has a greater improvement in accuracy than the traditional Quickshift algorithm and other clustering algorithms.