A color image retrieval algorithm was presented in order to retrieve image from massive images. This method firstly transformed image from RGB space to HSV space. V component of HSV space was decomposed into three levels by dual tree complex wavelet transform (DT-CWT) to extract high frequency components as texture features. Invariant moments of V component were extracted as shape features. Information of H component of HSV space was extracted as color features. Then three types of features were combined to form feature vector to judge similarity of images. Experiments show that this algorithm has not only high precision and recall but also low time consumption.