Flower image retrieval plays important role in automatic species recognation. In this paper we use segmentation based method on color clustering and domain knowledge to extract flower regions. Color Moments of flower region are used to characterize the color feature. In characterizing shape feature, we use centroid color distance (CCD) and angle code histogram (ACH). Experimental result showed that clustering based flower region extraction method can produce good accuracy. Flower retrieval result in 16 species of flower images where yyy species gave the maximum 89% accuracy. Keywords: Flower Image Retrieval, Flower Image Characterization, Region-Of-Interest (ROI), Color Features, Shape Features