This paper proposes a novel approach for clustering images based on their color by automatic fuzzy clustering algorithm of probability density functions. Herein, the difference between probability density functions derived from color matrix of images is evaluated by L1- distance. The proposed algorithm is compared with others in terms of various criteria such as misclustering rate, alternative Silhouette width, PBM index, adjusted rand index and computational time. The numerical examples show an outstanding performance of the proposed algorithm in comparison with that of existing algorithms and its potential applications in real problems.