Abstract: Camouflage color is the greatest factor in camouflage pattern with good background matching.An effective method for camouflage color selection based on improved K-means clustering is presented in this paper.Standard camouflage colors with great differences are selected as initial centroids，and pixels partition depends on color difference with cluster centroids and neighbors character，at last the background superior colors are transformed into standard camouflage colors ordained by army.Combining the above-mentioned theory with camouflage pattern design technique，the authors analyze the validity of the algorithm through experiments.The camouflage effect of camouflaged target in diverse backgrounds is also tested through edge detection and correlation tracking.The research result indicates that camouflage color selection based on improved K-means clustering is good enough for camouflage pattern design.