Traditional density-based clustering algorithm mainly has three problems as follow. Firstly, it only supports spatial
attributes without considering non-spatial attributes in the database. Secondly, it is difficult to set the parameters, and the
clustering result is sensitive to the parameters. Thirdly, it can’t discover the clusters of different density for adopting absolute
density as the metrics of all clusters. In order to overcome these problems mentioned above, the paper presents an relative
density-based clustering algorithm for mixture data sets(RDBC M), and further carries out the research on its incremental
clustering algorithm. Theoretical analysis and simulation experiment verfy the effectiveness and the performance speed-up
effect of the proposed algorithm.