论文标题
基于学习的图像压缩的二进制概率模型
Binary Probability Model for Learning Based Image Compression
论文作者
论文摘要
在本文中，我们建议使用潜在变量的概率模型来增强学习的图像压缩系统。先前的作品用高斯或拉普拉斯分布模型。受二进制算术编码的启发，我们建议用三个二进制值和一个整数，具有不同的概率模型来向潜伏期发出信号。放松方法旨在进行基于梯度的训练。更丰富的概率模型会导致更好的熵编码，从而降低速率。在学习的图像压缩（CLIC）测试条件下，在挑战下进行的实验表明，与高斯或拉普拉斯模型相比，该方法可节省18％的利率。
In this paper, we propose to enhance learned image compression systems with a richer probability model for the latent variables. Previous works model the latents with a Gaussian or a Laplace distribution. Inspired by binary arithmetic coding , we propose to signal the latents with three binary values and one integer, with different probability models. A relaxation method is designed to perform gradient-based training. The richer probability model results in a better entropy coding leading to lower rate. Experiments under the Challenge on Learned Image Compression (CLIC) test conditions demonstrate that this method achieves 18% rate saving compared to Gaussian or Laplace models.
