• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.3
• /
• pp.1700-1721
• /
• 2017

The problem of blocking artifacts is very common in block-based image and video compression, especially at very low bit rates. In this paper, we propose a post-processing method for JPEG-coded image deblocking via sparse representation and adaptive residual threshold. This method includes three steps. First, we obtain the dictionary by online dictionary learning and the compressed images. The dictionary is then modified by the histogram of oriented gradient (HOG) feature descriptor and K-means cluster. Second, an adaptive residual threshold for orthogonal matching pursuit (OMP) is proposed and used for sparse coding by combining blind image blocking assessment. At last, to take advantage of human visual system (HVS), the edge regions of the obtained deblocked image can be further modified by the edge regions of the compressed image. The experimental results show that our proposed method can keep the image more texture and edge information while reducing the image blocking artifacts.

• Journal of the Optical Society of Korea
• /
• v.20 no.6
• /
• pp.752-761
• /
• 2016

Hyperspectral images are often contaminated with stripe noise, which severely degrades the imaging quality and the precision of the subsequent processing. In this paper, a variational model is proposed by employing spectral-spatial adaptive unidirectional variation and a sparse representation. Unlike traditional methods, we exploit the spectral correction and remove stripes in different bands and different regions adaptively, instead of selecting parameters band by band. The regularization strength adapts to the spectrally varying stripe intensities and the spatially varying texture information. Spectral correlation is exploited via dictionary learning in the sparse representation framework to prevent spectral distortion. Moreover, the minimization problem, which contains two unsmooth and inseparable $l_1$-norm terms, is optimized by the split Bregman approach. Experimental results, on datasets from several imaging systems, demonstrate that the proposed method can remove stripe noise effectively and adaptively, as well as preserve original detail information.

• Journal of the Korea Computer Graphics Society
• /
• v.22 no.3
• /
• pp.21-29
• /
• 2016

Cell segmentation is an important but time-consuming and laborious task in biological image analysis. An automated, robust, and fast method is required to overcome such burdensome processes. These needs are, however, challenging due to various cell shapes, intensity, and incomplete boundaries. A precise cell segmentation will allow to making a pathological diagnosis of tissue samples. A vast body of literature exists on cell segmentation in microscopy images [1]. The majority of existing work is based on input images and predefined feature models only - for example, using a deformable model to extract edge boundaries in the image. Only a handful of recent methods employ data-driven approaches, such as supervised learning. In this paper, we propose a novel data-driven cell segmentation algorithm for bright-field microscopy images. The proposed method minimizes an energy formula defined by two dictionaries - one is for input images and the other is for their manual segmentation results - and a common sparse code, which aims to find the pixel-level classification by deploying the learned dictionaries on new images. In contrast to deformable models, we do not need to know a prior knowledge of objects. We also employed convolutional sparse coding and Alternating Direction of Multiplier Method (ADMM) for fast dictionary learning and energy minimization. Unlike an existing method [1], our method trains both dictionaries concurrently, and is implemented using the GPU device for faster performance.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.10 no.4
• /
• pp.257-267
• /
• 2017

Compressed Sensing(CS) deals with linear inverse problems. The theoretical results of CS have had an impact on inference problems and presented amazing research achievements in the related fields including signal processing and information theory. However, in order for CS to be applied in practical environments, there are two significant challenges to be solved. One is to guarantee in real time recovery of CS signals, and the other is that the signals have to be sparse. To this end, the latest researches using deep learning technology have emerged. In this paper, we consider CS problems based on deep learning and discuss the latest research results. And the approaches for CS signal reconstruction using deep learning show superior results in terms of recovery time and performance. It is expected that the approaches for CS reconstruction using deep learning shown in recent studies can not only raise the possibility of utilization of CS, but also be highly exploited in the fields of signal processing and communication areas.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2023.05a
• /
• pp.165-165
• /
• 2023

Streamflow prediction is a critical task in water resources management and essential for planning and decision-making purposes. However, the streamflow prediction is challenging due to the complexity and non-linear nature of hydrological processes. The transfer learning is a powerful technique that enables a model to transfer knowledge from a source domain to a target domain, improving model performance with limited data in the target domain. In this study, we apply the transfer learning using the Informer model, which is a state-of-the-art deep learning model for streamflow prediction. The model was trained on a large-scale hydrological dataset in the source basin and then fine-tuned using a smaller dataset available in the target basin to predict the streamflow in the target basin. The results demonstrate that transfer learning using the Informer model significantly outperforms the traditional machine learning models and even other deep learning models for streamflow prediction, especially when the target domain has limited data. Moreover, the results indicate the effectiveness of streamflow prediction when knowledge transfer is used to improve the generalizability of hydrologic models in data-sparse regions.

• Journal of Positioning, Navigation, and Timing
• /
• v.10 no.4
• /
• pp.297-305
• /
• 2021

Monocular depth estimation helps the robot to understand the surrounding environments in 3D. Especially, deep-learning-based monocular depth estimation has been widely researched, because it may overcome the scale ambiguity problem, which is a main issue in classical methods. Those learning based methods can be mainly divided into three parts: supervised learning, unsupervised learning, and semi-supervised learning. Supervised learning trains the network from dense ground-truth depth information, unsupervised one trains it from images sequences and semi-supervised one trains it from stereo images and sparse ground-truth depth. We describe the basics of each method, and then explain the recent research efforts to enhance the depth estimation performance.

• KNOM Review
• /
• v.24 no.2
• /
• pp.24-34
• /
• 2021

Service providers using edge computing provide a high level of service. As a result, devices store important information in inner storage and have become a target of the latest cyberattacks, which are more difficult to detect. Although experts use a security system such as intrusion detection systems, the existing intrusion systems have low detection accuracy. Therefore, in this paper, we proposed a machine learning model for more accurate intrusion detections of devices in edge computing. The proposed model is a hybrid model that combines a stacked sparse autoencoder (SSAE) and a convolutional neural network (CNN) to extract important feature vectors from the input data using sparsity constraints. To find the optimal model, we compared and analyzed the performance as adjusting the sparsity coefficient of SSAE. As a result, the model showed the highest accuracy as a 96.9% using the sparsity constraints. Therefore, the model showed the highest performance when model trains only important features.

• 제어로봇시스템학회:학술대회논문집
• /
• 2000.10a
• /
• pp.531-531
• /
• 2000

The challenge is to control unstable nonlinear dynamic systems using only sparse feedback from the environment concerning its performance. The design of such controllers can be achieved by evolving neural networks. An evolutionary approach to train neural networks in realtime is proposed. Evolutionary strategies adapt the weights of neural networks and the threshold values of neuron's synapses. The proposed method has been successfully implemented for pole balancing problem.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.10a
• /
• pp.9-11
• /
• 2022

Applying Reinforcement Learning in everyday applications and varied environments has proved the potential of the of the field and revealed pitfalls along the way. In robotics, a learning agent takes over gradually the control of a robot by abstracting the navigation model of the robot with its inputs and outputs, thus reducing the human intervention. The challenge for the agent is how to implement a feedback function that facilitates the learning process of an MDP problem in an environment while reducing the time of convergence for the method. In this paper we will implement a reward shaping system avoiding sparse rewards which gives fewer data for the learning agent in a ROS environment. Reward shaping prioritizes behaviours that brings the robot closer to the goal by giving intermediate rewards and helps the algorithm converge quickly. We will use a pseudocode implementation as an illustration of the method.

• Smart Structures and Systems
• /
• v.21 no.1
• /
• pp.1-14
• /
• 2018

A method is proposed to detect structural damages in the presence of damping using noisy data. This method uses Frequency Response Function (FRF) and Mode-Shapes as the input parameters for a system of Couple Sparse Coding (CSC) to study the healthy state of the structure. To obtain appropriate patterns of FRF for CSC training, Principal Component Analysis (PCA) technique is adopted to reduce the full-size FRF to overcome over-fitting and convergence problems in machine-learning training. To verify the proposed method, a numerical two-story frame structure is employed. A system of individual CSCs is trained with FRFs and mode-shapes, and then termed ensemble to detect the health condition of the structure. The results demonstrate that the proposed method is accurate in damage identification even in presence of up to 20% noisy data and 5% unconsidered damping ratio. Furthermore, it can be concluded that CSC ensemble is highly efficient to detect the location and the severity of damages in comparison to the individual CSC trained only with FRF data.