Research on Commodity Image Data Compression Based on SVD Algorithm

In recent years, with the development of E-commerce economy, growth of multimedia information data is very rapid. The data without image compression processing will be limited in processing and storage, and it is not conducive to the use of machine learning for data mining. In this paper, SVD (Singular Value Decomposition) image compression algorithm is used to reduce the dimension of the images in the commodity image data set of Harbin University of Commerce, and preserve the image features. This study mainly uses the scikit-learn open source machine learning library written in Python language, combined with the function of the module, to write the target algorithm. The experimental results show that different commodity images in the dataset can achieve 0.7-0.9 compression ratio through the algorithm, while maintaining good image features. It is significant to improve the efficiency of image data transmission and the machine learning research for this dataset.