• Journal of Information Processing Systems
• /
• v.9 no.1
• /
• pp.53-68
• /
• 2013

We investigate the rate-compatible punctured convolutional (RCPC) codes concatenated with hierarchical QAM for designing a cross-layer unequal error protection scheme for H.264 coded sequences. We first divide the H.264 encoded video slices into three priority classes based on their relative importance. We investigate the system constraints and propose an optimization formulation to compute the optimal parameters of the proposed system for the given source significance information. An upper bound to the significance-weighted bit error rate in the proposed system is derived as a function of system parameters, including the code rate and geometry of the constellation. An example is given with design rules for H.264 video communications and 3.5-4 dB PSNR improvement over existing RCPC based techniques for AWGN wireless channels is shown through simulations.

• Structural Monitoring and Maintenance
• /
• v.5 no.1
• /
• pp.51-65
• /
• 2018

A number of sensing techniques have been implemented for detecting defects in civil infrastructures instead of onsite human inspections in structural health monitoring. However, the issue of faults in sensors has not received much attention. This issue may lead to incorrect interpretation of data and false alarms. To overcome these challenges, this article presents a deep learning-based method with a new architecture of Stateful Long Short Term Memory Neural Networks (S-LSTM NN) for detecting sensor fault without going into details of the fault features. As LSTMs are capable of learning data features automatically, and the proposed method works without an accurate mathematical model. The detection of four types of sensor faults are studied in this paper. Non-stationary acceleration responses of a three-span continuous bridge when under operational conditions are studied. A deep network model is applied to the measured bridge data with estimation to detect the sensor fault. Another set of sensor output data is used to supervise the network parameters and backpropagation algorithm to fine tune the parameters to establish a deep self-coding network model. The response residuals between the true value and the predicted value of the deep S-LSTM network was statistically analyzed to determine the fault threshold of sensor. Experimental study with a cable-stayed bridge further indicated that the proposed method is robust in the detection of the sensor fault.

• Proceedings of the KIEE Conference
• /
• 2005.05a
• /
• pp.113-115
• /
• 2005

This thesis presents a method for 3D modeling of facial expression from frontal and mirror-reflected multi-image. Since the proposed system uses only one camera, two mirrors, and simple mirror's property, it is robust, accurate and inexpensive. In addition, we can avoid the problem of synchronization between data among different cameras. Mirrors located near one's cheeks can reflect the side views of markers on one's face. To optimize our system, we must select feature points of face intimately associated with human's emotions. Therefore we refer to the FDP (Facial Definition Parameters) and FAP (Facial Animation Parameters) defined by MPEG-4 SNHC (Synlhetic/Natural Hybrid Coding). We put colorful dot markers on selected feature points of face to detect movement of facial deformation when subject makes variety expressions. Before computing the 3D coordinates of extracted facial feature points, we properly grouped these points according to relative part. This makes our matching process automatically. We experiment on about twenty koreans the subject of our experiment in their late twenties and early thirties. Finally, we verify the performance of the proposed method tv simulating an animation of 3D facial expression.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.50 no.2
• /
• pp.185-192
• /
• 2013

In this paper, proposed to improve the performance of speech and mixed content signal classification using MFCC based on GMM probability model used for the MPEG USAC(Unified Speech and Audio Coding) standard. For effective pattern recognition, the Gaussian mixture model (GMM) probability model is used. For the optimal GMM parameter extraction, we use the expectation maximization (EM) algorithm. The proposed classification algorithm is divided into two significant parts. The first one extracts the optimal parameters for the GMM. The second distinguishes between speech and mixed content signals using MFCC feature parameters. The performance of the proposed classification algorithm shows better results compared to the conventionally implemented USAC scheme.

• ETRI Journal
• /
• v.42 no.2
• /
• pp.258-271
• /
• 2020

A correlation exists between luminance samples and chrominance samples of a color image. It is beneficial to exploit such interchannel redundancy for color image compression. We propose an algorithm that predicts chrominance components Cb and Cr from the luminance component Y. The prediction model is trained by supervised learning with Laplacian-regularized least squares to minimize the total prediction error. Kernel principal component analysis mapping, which reduces computational complexity, is implemented on the same point set at both the encoder and decoder to ensure that predictions are identical at both the ends without signaling extra location information. In addition, chrominance subsampling and entropy coding for model parameters are adopted to further reduce the bit rate. Finally, luminance information and model parameters are stored for image reconstruction. Experimental results show the performance superiority of the proposed algorithm over its predecessor and JPEG, and even over JPEG-XR. The compensation version with the chrominance difference of the proposed algorithm performs close to and even better than JPEG2000 in some cases.

• The Journal of the Acoustical Society of Korea
• /
• v.16 no.1
• /
• pp.67-75
• /
• 1997

In this paper, A method is proposed to analyze the noise figure for some parameters based on IS-98 recommendation when a RF receiver for the CDMA mobile station is designed. Simulation results show that the maximum noise figure of the receiver, which fits in the receiving sensitivity of the mobile station, is about 11dB, and this value is smaller about 3dB than the existing specification. In addition, we have tested the relationship between frame error rate and Eb/Nt of traffic channel for each coding rate. according to the speed changes of the traffic channel. In order to prove IMD to be most important variable in LNA we have tested IMD spurious in case that LNA always turns on and in case that LNA turns ON/OFF.

• ETRI Journal
• /
• v.31 no.5
• /
• pp.489-499
• /
• 2009

This paper focuses on assessment and design of transmission systems for distribution of digital signals over standard Category-7A copper cables at speeds beyond 10 Gbps. The main contribution of this paper is on the technical feasibility and system design for data rates of 40 Gbps and 100 Gbps over copper. Based on capacity analysis and rate optimization algorithms, system parameters are obtained and the design implementation trade-offs are discussed. Our simulation results confirm that with the aid of a decision-feedback equalizer and powerful coding techniques, namely, TCM or LDPC code, 40 Gbps transmission is feasible over 50 m of CAT-7A copper cable. These results also indicate that 100 Gbps transmission can be achieved over 15 m cables.

• 제어로봇시스템학회:학술대회논문집
• /
• 1997.10a
• /
• pp.120-123
• /
• 1997

This paper presents models of fuzzy inference systems that can be built from a set of input-output training data pairs through hybrid structure-parameter learning. Fuzzy inference systems has the difficulty of parameter learning. Here we develop a coding format to determine a fuzzy neural network(FNN) model by chromosome in a genetic algorithm(GA) and present systematic approach to identify the parameters and structure of FNN. The proposed FNN can automatically identify the fuzzy rules and tune the membership functions by modifying the connection weights of the networks using the GA and the back-propagation learning algorithm. In order to show effectiveness of it we simulate and compare with conventional methods.

• The Journal of the Acoustical Society of Korea
• /
• v.21 no.6
• /
• pp.531-542
• /
• 2002

Novel tandemless transcoding method for AMR and EVRC speech coders is proposed in this paper. In contrast to conventional tandem method, the parameters which is used commonly in speech coder where CELP algorithm is adapted are directly transcoded. The proposed algorithm is composed of LSP transcoding, pitch delay transcoding, gains transcoding and fixed codebook vector transcoding Evaluation results show that the novel algorithm achieves better speech quality than tandem method and reduce computational complexity and delay.

• Journal of Computational Design and Engineering
• /
• v.3 no.3
• /
• pp.274-285
• /
• 2016

A designer is mainly supported by two essential factors in design decisions. These two factors are intelligence and experience aiding the designer by predicting the interconnection between the required design parameters. Through classification of product data and similarity recognition between new and existing designs, it is partially possible to replace the required experience for an inexperienced designer. Given this context, the current paper addresses a framework for recognition and flexible retrieval of similar models in product design. The idea is to establish an infrastructure for transferring design as well as the required PLM (Product Lifecycle Management) know-how to the design phase of product development in order to reduce the design time. Furthermore, such a method can be applied as a brainstorming method for a new and creative product development as well. The proposed framework has been tested and benchmarked while showing promising results.