Recent advancements in deep learning-based image compression are notable. However, prevalent schemes that employ a serial context-adaptive entropy model to enhance rate-distortion (R-D) performance are markedly slow. Furthermore, the complexities of the encoding and decoding networks are substantially high, rendering them unsuitable for some practical applications. In this paper, we propose two techniques to balance the trade-off between complexity and performance. First, we introduce two branching coding networks to independently learn a low-resolution latent representation and a high-resolution latent representation of the input image, discriminatively representing the global and local information therein. Second, we utilize the high-resolution latent representation as conditional information for the low-resolution latent representation, furnishing it with global information, thus aiding in the reduction of redundancy between low-resolution information. We do not utilize any serial entropy models. Instead, we employ a parallel channel-wise auto-regressive entropy model for encoding and decoding low-resolution and high-resolution latent representations. Experiments demonstrate that our method is approximately twice as fast in both encoding and decoding compared to the parallelizable checkerboard context model, and it also achieves a 1.2% improvement in R-D performance compared to state-of-the-art learned image compression schemes. Our method also outperforms classical image codecs including H.266/VVC-intra (4:4:4) and some recent learned methods in rate-distortion performance, as validated by both PSNR and MS-SSIM metrics on the Kodak dataset.

Rate adaption is one of the decisive factors for the applications of video compression. However, previous deep video compression methods are usually optimized for a single fixed rate-distortion (R-D) tradeoff. While they can achieve multiple bitrates by training multiple independent models, the realized bitrates are limited to several discrete points on the R-D curve and the storage cost increases proportionally to the number of models. In this paper, we propose a variable-rate scheme for deep video compression, which can achieve continuously variable rate by a single model, i.e., it can reach any point on the R-D curve. In our scheme, two deep auto-encoders are used to compress the residual and the motion vector field respectively, which directly generate the final bitstream. The basic rate adaptation can be achieved by using the R-D tradeoff parameter to deeply modulate all the internal feature maps of the auto-encoders. However, other modules in our scheme, notably motion estimation and motion compensation, also affect the final bitrate indirectly. We further use the R-D tradeoff parameter to modulate them via a conditional map, which effectively improves the compression efficiency. We use a multi-rate-distortion loss function together with a step-by-step training strategy to optimize the entire scheme. Our experiments show that the proposed scheme achieves continuously variable rate by a single model with almost the same compression efficiency as multiple fixed-rate models. The additional parameters and computation of our model are negligible when compared with a single fixed-rate model.

Recently many works attempt to develop image compression models based on deep learning architectures, where the uniform scalar quantizer (SQ) is commonly applied to the feature maps between the encoder and decoder. In this paper, we propose to incorporate trellis coded quantizer (TCQ) into a deep learning based image compression framework. A soft-to-hard strategy is applied to allow for back propagation during training. We develop a simple image compression model that consists of three subnetworks (encoder, decoder and entropy estimation), and optimize all of the components in an end-to-end manner. We experiment on two high resolution image datasets and both show that our model can achieve superior performance at low bit rates. We also show the comparisons between TCQ and SQ based on our proposed baseline model and demonstrate the advantage of TCQ.

The graph-based transform has recently emerged as an effective tool for compressing some special signals such as depth images in 3D videos. However, one limitation of this approach is that it needs to apply eigen-decomposition to each block. To reduce the complexity, in this paper, we develop a systematic approach to find a universal optimal graph-based transform for a class of 2D piecewise smooth signals. Each block in the class can include a discontinuity whose locations in different rows are randomly distributed within a confined region. We first define a special 2D graph model for this class of signals. Our derivation then reveals that the inverse of the covariance matrix of this class of signals is equal to its graph Laplacian with a bias value added to the first diagonal element. Furthermore, the edge values within the confined region have a closed-form expression. If the bias value is assumed negligible then the approximation of the optimal transform for the class of signals is given by the eigenvectors of the true graph Laplacian and can be pre-computed. Therefore online eigen-decomposition for the class of signals can be avoided, and the complexity of the encoder and decoder can thus be reduced. The feasibility of the proposed scheme is demonstrated via depth image coding examples.

A low complexity multiple description (MD) coding method is proposed to generate M descriptions. Consider the MD coding of a stationary correlated source. We first fictitiously partition the source into sample blocks of size M, i.e., M polyphases. Each description encodes all input samples, but with a variable bit rate that depends on the indices of the sample and the description. A special DPCM encoder is used in each description, where each sample is predicted from the reconstructed samples in the same description. The prediction error is uniformly scalar-quantized and entropy coded.

This paper investigates the design and application of the optimal filter banks for a prediction-compensated multiple description coding (PC-MDC) scheme, where the coefficients in each subband are split into two descriptions. Each description also includes the prediction residuals of the data in the other description. The optimal designs of orthogonal and biorthogonal filter banks with multiple-level decompositions are formulated in a unified framework. The optimal results in all cases are found to be very close to the optimal filter banks in traditional single description coding. This allows us to apply the proposed method to existing systems with single-description-optimized filter banks and still enjoy near-optimal performance. Image coding results in the JPEG 2000 framework show that the proposed method achieves similar or better performance than other methods. It also has lower complexity and is more compatible to the JPEG 2000 standard.

This paper presents a scheme to design directional lapped transforms. Lapped transforms can be factorized into lifting steps. By introducing directional operator into each lifting step, the directional lapped transform is constructed. The directional lapped transform proposed not only preserves the advantages of lapped transforms, it also can represent directional signals more efficiently. An image coding scheme using the directional lapped transform is also described. Compared to the state-of-the-art image coding using lapped transform, HD photo, the proposed scheme shows more than 20 dB's gain for artificial images with strong directional correlations. And for natural images, up to 1.5 dB's gain can also be observed.