Abstract: Based on the introduction of image entropy into the den sity peaks clustering algorithm to determine the information quantity of spectra l band, we proposed a method for density peaks clustering band selection based o n image entropy. By constructing the measuring score of bands signif icance， we solved the problem of the band selection in the traditional density peaks clustering method without considering the info rmation quantity of the bands. The experiment al results show that the classification accuracy of the proposed method increase s by an average of 2.12% than that of the traditional density peaks clustering method.

Key words: band selection (BS), hyperspectral remote sensing, density peak clustering (DPC), image entropy