Channel-Level Variable Quantization Network for Deep Image Compression
This work addresses a bottleneck in deep image compression for applications requiring efficient storage and transmission, though it is incremental as it builds on existing joint rate-distortion frameworks.
The paper tackles the inflexibility of treating all channels equally in deep image compression by proposing a channel-level variable quantization network that dynamically allocates bitrates based on channel importance, resulting in superior performance and better visual reconstructions.
Deep image compression systems mainly contain four components: encoder, quantizer, entropy model, and decoder. To optimize these four components, a joint rate-distortion framework was proposed, and many deep neural network-based methods achieved great success in image compression. However, almost all convolutional neural network-based methods treat channel-wise feature maps equally, reducing the flexibility in handling different types of information. In this paper, we propose a channel-level variable quantization network to dynamically allocate more bitrates for significant channels and withdraw bitrates for negligible channels. Specifically, we propose a variable quantization controller. It consists of two key components: the channel importance module, which can dynamically learn the importance of channels during training, and the splitting-merging module, which can allocate different bitrates for different channels. We also formulate the quantizer into a Gaussian mixture model manner. Quantitative and qualitative experiments verify the effectiveness of the proposed model and demonstrate that our method achieves superior performance and can produce much better visual reconstructions.