Medical Image Compression Based on Variational Autoencoder

With the rapid growth of medical image data, it has become a current research hotspot that how to realize large amounts real-time upload and storage images with limited network bandwidth space. However, currently, compression technology cannot perform joint optimization rate (the degree compression) distortion (reconstruction effect). Therefore, this study proposed algorithm based on variational autoencoder. This takes as common goal uses residual module directly transmit information, which alleviates contradiction between improving optimizing reconstruction effect. At same time, also reduces loss in process by adding network. The experimental results show that, compared traditional deep learning algorithm, smaller distortion, better effect, can obtain higher quality at rate.