Abstract: Aiming at the defects of single feature that could only describe the partial information of image content. In order to obtain more accurate image retrieval results, we designed an image retrieval algorithm based on multifeature combination and user feedback. Firstly, many features of the image were extracted and combined to form a feature vector for image retrieval, the Euclidean distance was used to calculate the similarity between images, and the
 initial image retrieval was realized. Secondly, through the  evaluation and feedback of the initial image retrieval results, the machine learning algorithm was used to adjust the initial image retrieval results according to the feedback results, so as to achieve accurate image retrieval. Finally, the standard data of image library retrieval were simulated and tested. The test results show that the proposed algorithm achieves stable image retrieval results, and the accuracy of image retrieval is obviously higher than other algorithms.

Key words: user feedback, feature vector, retrieval accuracy, machine learning algorithm, image retrieval