Multimodal learning, especially large-scale multimodal pre-training, has developed rapidly over the past few years and led to the greatest advances in artificial intelligence (AI). Despite its effectiveness, understan...
详细信息
In grey-level image restoration, a prior knowledge of degraded areas allows, thanks to the selective filtering, to achieve a good protection of the image features. In this paper, we propose a quadratic programming-bas...
详细信息
In grey-level image restoration, a prior knowledge of degraded areas allows, thanks to the selective filtering, to achieve a good protection of the image features. In this paper, we propose a quadratic programming-based technique that deals with the issue of details preservation during the restoration process. Based on the classical model of image restoration, we build a modified model by introducing a set of binary variables that indicate the pixel categories. We combine each pixel with the median of its neighbours in a decision rule so that one of them generates the optimal solution. The obtained model is a nonlinear mixed-integer problem where resolution by exact methods is not feasible. In this regard, we use both of the continuous Hopfield neural network and the genetic algorithm to solve the suggested model. Performance of our method is demonstrated numerically and visually by several computational tests.
In this paper, a general variational probabilistic generative framework parameterized by deep networks is proposed for single image super-resolution, which assembles the advantages of coding-based methods and regressi...
详细信息
In this paper, a general variational probabilistic generative framework parameterized by deep networks is proposed for single image super-resolution, which assembles the advantages of coding-based methods and regression-based methods. We use probabilistic generative networks to model the joint full likelihood of a pair of low-resolution (LR) and high-resolution (HR) patches which are generated from a shared latent representation. An inference model is applied to infer the stochastic distribution of the latent representation. By jointly optimizing the generative and inference models, a regression process to the distribution of the HR patch is implied during the learning phase, which provides an efficient forward mapping to accomplish the super-resolution task. We use our framework as a guidance and develop a new model called PGM-CP, with the help of an informative conditional prior and a consistent recognition model. We likewise show how three existing popular example-based SR methods can be "reinvented" under our framework. The effectiveness and efficiency of the proposed method is examined based on three public datasets. Experimental results demonstrate that our model is competitive with state-of-the-art approaches, especially when the image is corrupted by noise. (C) 2018 Elsevier B.V. All rights reserved.
In this work recent advances in conditional adversarial networks are investigated to develop an end-to-end architecture based on Convolutional neural Networks (CNNs) to directly map realistic colours to an input greys...
详细信息
ISBN:
(纸本)9781728118178
In this work recent advances in conditional adversarial networks are investigated to develop an end-to-end architecture based on Convolutional neural Networks (CNNs) to directly map realistic colours to an input greyscale image. Observing that existing colourisation methods sometimes exhibit a lack of colourfulness, this paper proposes a method to improve colourisation results. In particular, the method uses Generative Adversarial neural Networks (GANs) and focuses on improvement of training stability to enable better generalisation in large multi-class image datasets. Additionally, the integration of instance and batch normalisation layers in both generator and discriminator is introduced to the popular U-Net architecture, boosting the network capabilities to generalise the style changes of the content. The method has been tested using the ILSVRC 2012 dataset, achieving improved automatic colourisation results compared to other methods based on GANs.
Skeletal muscles are functionally regulated by populations of so-called motor units (MUs). An MU comprises a bundle of muscle fibers controlled by a neuron from the spinal cord. Current methods to diagnose neuromuscul...
详细信息
Skeletal muscles are functionally regulated by populations of so-called motor units (MUs). An MU comprises a bundle of muscle fibers controlled by a neuron from the spinal cord. Current methods to diagnose neuromuscular diseases and monitor rehabilitation, and study sports sciences rely on recording and analyzing the bio-electric activity of the MUs. However, these methods provide information from a limited part of a muscle. Ultrasound imaging provides information from a large part of the muscle. It has recently been shown that ultrafast ultrasound imaging can be used to record and analyze the mechanical response of individual MUs using blind source separation. In this work, we present an alternative method - a deep learning pipeline - to identify active MUs in ultrasound image sequences, including segmentation of their territories and signal estimation of their mechanical responses (twitch train). We train and evaluate the model using simulated data mimicking the complex activation pattern of tens of activated MUs with overlapping territories and partially synchronized activation patterns. Using a slow fusion approach (based on 3D CNNs), we transform the spatiotemporal image sequence data to 2D representations and apply a deep neural network architecture for segmentation. Next, we employ a second deep neural network architecture for signal estimation. The results show that the proposed pipeline can effectively identify individual MUs, estimate their territories, and estimate their twitch train signal at low contraction forces. The framework can retain spatio-temporal consistencies and information of the mechanical response of MU activity even when the ultrasound image sequences are transformed into a 2D representation for compatibility with more traditional computer vision and imageprocessing techniques. The proposed pipeline is potentially useful to identify simultaneously active MUs in whole muscles in ultrasound image sequences of voluntary skeletal muscle cont
Despite extensive research efforts, blind image deblurring remains a challenge without general robust solutions. A longoverlooked problem of existing deblurring methods is that they are all designed to work on fully s...
详细信息
ISBN:
(纸本)9781538662496
Despite extensive research efforts, blind image deblurring remains a challenge without general robust solutions. A longoverlooked problem of existing deblurring methods is that they are all designed to work on fully sampled RGB input images for simplicity. But, in practice, most RGB color images are reconstructed from Bayer mosaic data hence riddled with various high-frequency demosaicking artifacts, such as zippering and moir ' e patterns, which can easily derail a deblurring algorithm. In this paper, we propose a novel multiscale deep convolutional neural network to solve demosaicking and deblurring jointly. By processing Bayer raw images directly, our method is free of the interference of demosaicking artifacts. Extensive experiments show that the joint approach greatly outperforms the simple cascade of state-of-art demosaicking and deblurring methods.
Multi-modal image acquisition techniques have allowed digital images to penetrate domains from micro-scale medical imaging to mega-scale satellite imaging. For postprocessing, deep learning techniques have widely been...
详细信息
Multi-modal image acquisition techniques have allowed digital images to penetrate domains from micro-scale medical imaging to mega-scale satellite imaging. For postprocessing, deep learning techniques have widely been used for image denoising and artifact suppression. However, far little work has been done to summarize their effectiveness concerning adaptive filter design, e.g., salt and pepper noise, stochastic Poisson, or additive white noise. Because different images, natural or urban, structured or unstructured scenes, and objects produce different types of noise, from the modality as well as from the imaging medium, devising a single method for all noise types is impractical. This paper proposes to use reinforcement learning (Q-learning) to adaptively design filters of a convolutional neural network (CNN). In contrast to the popular state of the art methods that use filter designs based on the noise model, CNN filters lack the power to do so. We have attempted to address this limitation of CNN by introducing a new modality of reinforcement learning for adaptive filter design. The qualitative and quantitative analysis of the proposed method is done and its efficacy is demonstrated using the following evaluation metrics: Peak signal to Noise Ratio (PSNR), Contrast to Noise Ratio (CNR), and Structure Similarity Index (SSIM).
A lot of important disease information can be accessed by performing retinal blood vessel analysis on fundus images. Diabetic retinopathy is one of the diseases understood by retinal blood vessel analysis. If this dis...
详细信息
A lot of important disease information can be accessed by performing retinal blood vessel analysis on fundus images. Diabetic retinopathy is one of the diseases understood by retinal blood vessel analysis. If this disease is detected at an early stage, vision loss can be prevented. In this paper, a method that performs retinal blood vessel analysis with classical methods is proposed. In this proposed system, pixel-based feature extraction is performed. Five different feature groups are used for feature extraction. These feature groups are edge detection, morphological, statistical, gradient, and Hessian matrix. An 18-D feature vector is created for each pixel. This feature vector is given to the artificial neural network for training. Using test images, the system is tested on two publicly available datasets. Sensitivity, Specificity, and Accuracy performance measures were used as success measures. The similarity index between the segmented image and the ground truth is measure using Dice and Jaccard. The accuracy of the system was measured as 96.18% for DRIVE and 94.56% for STARE, respectively. Experimental results show that the proposed algorithm achieves satisfactory results. This method can be used as an automated retinal blood vessel segmenting system.
Face spoofing detection is gaining an increasing attention in the biometric research. Various approaches have been proposed in the literatures. In these methods, the color variation of facial regions, caused by the de...
详细信息
暂无评论