Improved hybrid layered image compression using deep learning and traditional codecs

Recently deep learning-based methods have been applied in image compression and achieved many promising results. In this paper, we propose an improved hybrid layered image compression framework by combining deep learning and the traditional image codecs. At the encoder, we first use a convolutional neural network (CNN) to obtain a compact representation of the input image, which is losslessly encoded by the FLIF codec as the base layer of the bit stream. A coarse reconstruction of the input is obtained by another CNN from the reconstructed compact representation. The residual between the input and the coarse reconstruction is then obtained and encoded by the H.265/HEVC-based BPG codec as the enhancement layer of the bit stream. Experimental results using the Kodak and Tecnick datasets show that the proposed scheme outperforms the state-of-the-art deep learning-based layered coding scheme and traditional codecs including BPG in both PSNR and MS-SSIM metrics across a wide range of bit rates, when the images are coded in the RGB444 domain.     

Visual-PSNR measure of image quality

Objective assessment of image quality is important in numerous image and video processing applications. Many objective measures of image quality have been developed for this purpose, of which peak signal-to-noise ratio PSNR is one of the simplest and commonly used. However, it sometimes does not match well with objective mean opinion scores (MOS). This paper presents a novel objective full-reference measure of image quality (VPSNR), which is a modified PSNR measure. It will be shown that VPSNR takes into account some features of the human visual system (HVS). The performance of VPSNR is validated using a data set of four image databases, and in this article it is shown that for images compressed by block-based compression algorithms (like JPEG) the proposed measure in the pixel domain matches well with MOS.     

Learning to predict the quality of distorted-then-compressed images via a deep neural network

Being captured by amateur photographers, reciprocally propagated through multimedia pipelines, and compressed with different levels, real-world images usually suffer from a wide variety of hybrid distortions. Faced with this scenario, full-reference (FR) image quality assessment (IQA) algorithms can not deliver promising predictions due to the inferior references. Meanwhile, existing no-reference (NR) IQA algorithms remain limited in their efficacy to deal with different distortion types. To address this obstacle, we explore a NR-IQA metric by predicting the perceptual quality of distorted-then-compressed images using a deep neural network (DNN). First, we propose a novel two-stream DNN to handle both authentic distortions and synthetic compressions and adopt effective strategies to pre-train the two branches of the network. Specifically, we transfer the knowledge learned from in-the-wild images to account for authentic distortions by utilizing a pre-trained deep convolutional neural network (CNN) to provide meaningful initializations. Meanwhile, we build a CNN for synthetic compressions and pre-train it on a dataset including synthetic compressed images. Subsequently, we bilinearly pool these two sets of features as the image representation. The overall network is fine-tuned on an elaborately-designed auxiliary dataset, which is annotated by a reliable objective quality metric. Furthermore, we integrate the output of the authentic-distortion-aware branch with that of the overall network following a two-step prediction manner to boost the prediction performance, which can be applied in the distorted-then-compressed scenario when the reference image is available. Extensive experimental results on several databases especially on the LIVE Wild Compressed Picture Quality Database show that the proposed method achieves state-of-the-art performance with good generalizability and moderate computational complexity.     

Non-reference image quality assessment based on deep clustering

Image quality assessment (IQA) is an indispensable technique in computer vision and pattern recognition Existing deep IQA methods have achieved remarkable performance. As far as we know, these deep learning-based IQA algorithms lack an adaptive features extraction mechanism toward input images with varying sizes and the stability in avoiding disturbance from data noises and model deviation. To solve these problems, we propose a non-reference IQA method by designing a novel unsupervised deep clustering framework, where a 13-layer network structure is proposed that upgrades the fully-connected layers to produce high-level features with adaptive sizes. Moreover, we add a contracted regular term with a contracted autoencoder into the clustering loss function to form a quality model reflecting the clustering structure. Compared to the other IQA algorithms, our model with simple structure exhibits more stable and robust performance by the initial configuration of network parameters during end-to-end training. The experimental results on the LIVE and CSIP databases have shown that our method not only performs better than the state-of-the-art IQA algorithms, but also has a simpler structure and better adaptability.     

Full-reference IPTV image quality assessment by deeply learning structural cues

Image quality assessment (IQA) attempts to quantify the quality-aware visual attributes perceived by humans. They can be divided into subjective and objective image quality assessment. Subjective IQA algorithms rely on human judgment of image quality, where the human visual perception functions as the dominant factor However, they cannot be widely applied in practice due to the heavy reliance on different individuals. Motivated by the fact that objective IQA largely depends on image structural information, we propose a structural cues-based full-reference IPTV IQA algorithm. More specifically, we first design a grid-based object detection module to extract multiple structural information from both the reference IPTV image (i.e., video frame) and the test one. Afterwards, we propose a structure-preserved deep neural networks to generate the deep representation for each IPTV image. Subsequently, a new distance metric is proposed to measure the similarity between the reference image and the evaluated image. A test IPV image with a small calculated distance is considered as a high quality one. Comprehensive comparative study with the state-of-the-art IQA algorithms have shown that our method is accurate and robust.     

Learning-based image restoration for compressed images

In this paper, we propose a novel learning-based image restoration scheme for compressed images by suppressing compression artifacts and recovering high frequency (HF) components based upon the priors learnt from a training set of natural images. The JPEG compression process is simulated by a degradation model, represented by the signal attenuation and the Gaussian noise addition process. Based on the degradation model, the input image is locally filtered to remove Gaussian noise. Subsequently, the learning-based restoration algorithm reproduces the HF component to handle the attenuation process. Specifically, a Markov-chain based mapping strategy is employed to generate the HF primitives based on the learnt codebook. Finally, a quantization constraint algorithm regularizes the reconstructed image coefficients within a reasonable range, to prevent possible over-smoothing and thus ameliorate the image quality. Experimental results have demonstrated that the proposed scheme can reproduce higher quality images in terms of both objective and subjective quality.     

Content-weighted mean-squared error for quality assessment of compressed images

Image quality assessment (IQA) has been intensively studied, especially for the full-reference (FR) scenario. However, only the mean-squared error (MSE) is widely employed in compression. Why other IQA metrics work ineffectively? We first sum up three main limitations including the computational time, portability, and working manner. To address these problems, we then in this paper propose a new content-weighted MSE (CW-MSE) method to assess the quality of compressed images. The design principle of our model is to use adaptive Gaussian convolution to estimate the influence of image content in a block-based manner, thereby to approximate the human visual perception to image quality. Results of experiments on six popular subjective image quality databases (including LIVE, TID2008, CSIQ, IVC, Toyama and TID2013) confirm the superiority of our CW-MSE over state-of-the-art FR IQA approaches.     

An optimized CNN-based quality assessment model for screen content image

Most existing convolutional neural network (CNN) based models designed for natural image quality assessment (IQA) employ image patches as training samples for data augmentation, and obtain final quality score by averaging all predicted scores of image patches. This brings two problems when applying these methods for screen content image (SCI) quality assessment. Firstly, SCI contains more complex content compared to natural image. As a result, qualities of SCI patches are different, and the subjective differential mean opinion score (DMOS) is not appropriate as qualities of all image patches. Secondly, the average score of image patches does not represent the quality of entire SCI since the human visual system (HVS) is sensitive to image patches containing texture and edge information. In this paper, we propose a novel quadratic optimized model based on the deep convolutional neural network (QODCNN) for full-reference (FR) and no-reference (NR) SCI quality assessment to overcome these two problems. The contribution of our algorithm can be concluded as follows: 1) Considering the characteristics of SCIs, a valid network architecture is designed for both NR and FR visual quality evaluation of SCIs, which makes the networks learn the feature differences for FR-IQA; 2) with the consideration of correlation between local quality and DMOS, a training data selection method is proposed to fine-tune the pre-trained model with valid SCI patches; 3) an adaptive pooling approach is employed to fuse patch quality to obtain image quality, owns strong noise robust and effects on both FR and NR IQA. Experimental results verify that our model outperforms both current no-reference and full-reference image quality assessment methods on the benchmark screen content image quality assessment database (SIQAD). Cross-database evaluation shows high generalization ability and high effectiveness of our model.     

双分支残差去马赛克网络

为了缓解传统拜耳型去马赛算法中常出现的拉链和伪影等问题,提出一个新颖的基于深度学习的去马赛克算法。所提算法首先对马赛克图像中的红色、绿色及蓝色通道中的像素进行分解、剔除及组合等操作得到两幅彩色图像,然后将这两幅彩色图像输入到设计的卷积神经网络中,以重建出完整的彩色图像,该网络能充分地利用卷积层所生成的特征信息。实验结果表明,所提算法重建出的完整彩色图像的质量相对较高,并且在一定程度上缓解了拉链和伪影等问题,其客观指标和主观评价都优于对比算法。     

Speckle noise removal based on structural convolutional neural networks with feature fusion for medical image

Application of convolutional neural networks (CNNs) for image additive white Gaussian noise (AWGN) removal has attracted considerable attentions with the rapid development of deep learning in recent years. However, the work of image multiplicative speckle noise removal is rarely done. Moreover, most of the existing speckle noise removal algorithms are based on traditional methods with human priori knowledge, which means that the parameters of the algorithms need to be set manually. Nowadays, deep learning methods show clear advantages on image feature extraction. Multiplicative speckle noise is very common in real life images, especially in medical images. In this paper, a novel neural network structure is proposed to recover noisy images with speckle noise. Our proposed method mainly consists of three subnetworks. One network is rough clean image estimate subnetwork. Another is subnetwork of noise estimation. The last one is an information fusion network based on U-Net and several convolutional layers. Different from the existing speckle denoising model based on the statistics of images, the proposed network model can handle speckle denoising of different noise levels with an end-to-end trainable model. Extensive experimental results on several test datasets clearly demonstrate the superior performance of our proposed network over state-of-the-arts in terms of quantitative metrics and visual quality.     

Screen content image quality assessment based on convolutional neural networks

Screen content image (SCI) is a composite image including textual and pictorial regions resulting in many difficulties in image quality assessment (IQA). Large SCIs are divided into image patches to increase training samples for CNN training of IQA model, and this brings two problems: (1) local quality of each image patch is not equal to subjective differential mean opinion score (DMOS) of an entire image; (2) importance of different image patches is not same for quality assessment. In this paper, we propose a novel no-reference (NR) IQA model based on the convolutional neural network (CNN) for assessing the perceptual quality of SCIs. Our model conducts two designs solving problems which benefits from two strategies. For the first strategy, to imitate full-reference (FR) CNN-based model behavior, a CNN-based model is designed for both FR and NR IQA, and performance of NR-IQA part improves when the image patch scores predicted by FR-IQA part are adopted as the ground-truth to train NR-IQA part. For the second strategy, image patch qualities of one entire SCI are fused to obtain the SCI quality with an adaptive weighting method taking account the effect of the different image patch contents. Experimental results verify that our model outperforms all test NR IQA methods and most FR IQA methods on the screen content image quality assessment database (SIQAD). On the cross-database evaluation, the proposed method outperforms the existing NR IQA method in terms of at least 2.4 percent in PLCC and 2.8 percent in SRCC, which shows high generalization ability and high effectiveness of our model.     

Noncausal predictive image codec

The paper describes a lossy image codec that uses a noncausal (or bilateral) prediction model coupled with vector quantization. The noncausal prediction model is an alternative to the causal (or unilateral) model that is commonly used in differential pulse code modulation (DPCM) and other codecs with a predictive component. We show how to obtain a recursive implementation of the noncausal image model without compromising its optimality and how to apply this in coding in much the same way as a causal predictor. We report experimental compression results that demonstrate the superiority of using a noncausal model based predictor over using traditional causal predictors. The codec is shown to produce high-quality compressed images at low bit rates such as 0.375 b/pixel. This quality is contrasted with the degraded images that are produced at the same bit rates by codecs using causal predictors or standard discrete cosine transform/Joint Photographic Experts Group-based (DCT/JPEG-based) algorithms.     

High-quality face image generation based on generative adversarial networks

Conventional face image generation using generative adversarial networks (GAN) is limited by the quality of generated images since generator and discriminator use the same backpropagation network. In this paper, we discuss algorithms that can improve the quality of generated images, that is, high-quality face image generation. In order to achieve stability of network, we replace MLP with convolutional neural network (CNN) and remove pooling layers. We conduct comprehensive experiments on LFW, CelebA datasets and experimental results show the effectiveness of our proposed method.     

A weighted full-reference image quality assessment based on visual saliency

In full reference image quality assessment (IQA), the images without distortion are usually employed as reference, while the structures in both reference images and distorted images are ignored and all pixels are equally treated. In addition, the role of human visual system (HVS) is not taken account into subjective IQA metric. In this paper, a weighted full-reference image quality metric is proposed, where a weight imposed on each pixel indicates its importance in IQA. Furthermore, the weights can be estimated via visual saliency computation, which can approximate the subjective IQA via exploiting the HVS. In the experiments, the proposed metric is compared with several objective IQA metrics on LIVE release 2 and TID 2008 database. The results demonstrate that SROCC and PLCC of the proposed metric are 0.9647 and 0.9721, respectively,which are higher than other methods and it only takes 427.5 s, which is lower than that of most other methods.     

Deep quality assessment toward defogged aerial images

In order to improve the visual appearance of defogged of aerial images, in this work, a novel defogging algorithm based on conditional generative adversarial network is proposed. More specifically, the training process is carried out through an end-to-end trainable deep neural network. In detail, we upgrade the traditional adversarial loss function by incorporating an L1-regularized gradient to encode a rich set of detailed visual information inside each aerial image. In practice, to our best knowledge, existing image quality assessment algorithms might have deviation and supersaturation distortion on aerial images. To alleviate this problem, we leverage a random forest classification model to learn the mapping relationship between aerial image features and the quality ranking results. Subsequently, we transform the objective of defogged image quality assessment into a classification problem. Comprehensive experimental results on our compiled fogged aerial images quality data set have clearly demonstrated the effectiveness of our proposed algorithm.     

一个图像质量盲评估的统计测度

本文提出了一个图像质量盲评估的统计测度.该测度首先根据自然图像的统计性质与失真图像的模型,实现对图像小波系数分布参数的盲估计;再利用估计的分布参数来计算失真图像与参考图像之间的互信息,以量化失真图像对参考图像的保真度,进而实现对图像质量的评估.本文提出的测度避免了对参考图像的依赖,且克服了现有图像质量盲评估对特征选择与提取、机器学习等过程的依赖.LIVE图像质量评估数据库的总体评估结果表明：本文提出的盲评估统计测度对图像质量评估结果与数据库的主观评估结果高度一致,且优于文献中报道的盲评估测度.     

An ensemble image quality assessment algorithm based on deep feature clustering

To effective handle image quality assessment (IQA) where the images might be with sophisticated characteristics, we proposed a deep clustering-based ensemble approach for image quality assessment toward diverse images. Our approach is based on a convolutional DAE-aware deep architecture. By leveraging a layer-by-layer pre-training, our proposed deep feature clustering architecture extracted a fixed number of high-level features at first. Then, it optimally splits image samples into different clusters by using the fuzzy C-means algorithm based on the engineered deep features. For each cluster, we simulated a particular fitting function of differential mean opinion scores with each assessed image’s PSNR, SIMM, and VIF scores. Comprehensive experimental results on TID2008, TID2013 and LIVE databases have demonstrated that compared to the state-of-the-art counterparts, our proposed IQA method can reflect the subjective quality of images more accurately by seamlessly integrating the advantages of three existed IQA methods.     

BINet: A binary inpainting network for deep patch-based image compression

Recent deep learning models outperform standard lossy image compression codecs. However, applying these models on a patch-by-patch basis requires that each image patch be encoded and decoded independently. The influence from adjacent patches is therefore lost, leading to block artefacts at low bitrates. We propose the Binary Inpainting Network (BINet), an autoencoder framework which incorporates binary inpainting to reinstate interdependencies between adjacent patches, for improved patch-based compression of still images. When decoding a patch, BINet additionally uses the binarised encodings from surrounding patches to guide its reconstruction. In contrast to sequential inpainting methods where patches are decoded based on previous reconstructions, BINet operates directly on the binary codes of surrounding patches without access to the original or reconstructed image data. Encoding and decoding can therefore be performed in parallel. We demonstrate that BINet improves the compression quality of a competitive deep image codec across a range of compression levels.