A Comparative Study of Image Features and Similarity Measurement Methods in Cross-modal Retrieval of Commodity Images

With the rapid development of information technology and the widespread popularity of mobile devices and e-commerce trading platforms, the number and scale of commodity images and corresponding description text and other multimedia data are growing rapidly, and the demand for cross-modal retrieval of commodity images is also increasing. Cross- mode hash method has been widely studied and applied. Based on the matrix three-factor decomposition hash cross mode retrieval method, the problem of mapping hash code with different modes is solved. In the cross-modal retrieval process, in order to study and analyze different similarity measurement methods, when searching for product image cross-modalities, in order to analyze the retrieval results of different similarity measurement methods, the potential characteristics of different images are The effect of the image is analyzed and the performance of cross-mode retrieval under the matrix three-factor decomposition hash framework is compared. The research results show the results of cross-modal retrieval of commodity images with different similarity measurement methods and different image underlying features.