The design of a clustering algorithm based on density and distribution trend

The 2D space in which sample data are distributed is divided into a series of basic logic grids. The information of density and distribution trend of samples in each grid is acquired and used to cluster the samples by merging adjacent grids according to specified criteria. For every clustering related application, the parameters in this algorithm, such as the size of the grid, the threshold of density and distribution trend, can be preset and adjusted according to specific applications. Experimental results show that this algorithm is fast and efficient.