We report for the first time the theoretical and experimental results of the application of fuzzy sets theory in color-structure recognition.In this application,Karhunen-Loeve(K-L) transformations are used to transform color images in RGB color space into that in I 1,I 2,I 3 color space,where histograms of the color images are used as fuzzy sets and their membership functions are used to extract the characteristics of the color images.In this way,similarity principle of fuzzy sets theory can be employed in the recognition of color-structure of color images.Our theoretical analysis and experimental results show clearly that the results will not be affected by the subjective discrimination and the speed of the recognition will be increased obviously.