• Journal of information and communication convergence engineering
• /
• v.11 no.3
• /
• pp.207-215
• /
• 2013

In this paper, we address the challenging computer vision problem of obtaining a reliable facial expression analysis from a naturally interacting person. We propose a system that combines a 3D generic face model, 3D head tracking, and 2D tracker to track facial landmarks and recognize expressions. First, we extract facial landmarks from a neutral frontal face, and then we deform a 3D generic face to fit the input face. Next, we use our real-time 3D head tracking module to track a person's head in 3D and predict facial landmark positions in 2D using the projection from the updated 3D face model. Finally, we use tracked 2D landmarks to update the 3D landmarks. This integrated tracking loop enables efficient tracking of the non-rigid parts of a face in the presence of large 3D head motion. We conducted experiments for facial expression recognition using both framebased and sequence-based approaches. Our method provides a 75.9% recognition rate in 8 subjects with 7 key expressions. Our approach provides a considerable step forward toward new applications including human-computer interactions, behavioral science, robotics, and game applications.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.19 no.5
• /
• pp.655-660
• /
• 2009

The context prediction algorithms are not suitable to provide real-time personalized service for users in context-awareness environment. The algorithms have problems like time delay in training data processing and the difficulties of implementation in real-time environment. In this paper, we propose a prediction algorithm with user modeling to shorten of processing time and to improve the prediction accuracy in the context prediction algorithm. The algorithm uses moving path of user contexts for context prediction and generates user model by time-series analysis of user's moving path. And that predicts the user context with the user model by sequence matching method. We compared our algorithms with the prediction algorithms by processing time and prediction accuracy. As the result, the prediction accuracy of our algorithm is similar to the prediction algorithms, and processing time is reduced by 40% in real time service environment.

• KIPS Transactions on Software and Data Engineering
• /
• v.6 no.4
• /
• pp.203-210
• /
• 2017

In this paper, we propose an effective neural network model for image caption generation and model transfer. This model is a kind of multi-modal recurrent neural network models. It consists of five distinct layers: a convolution neural network layer for extracting visual information from images, an embedding layer for converting each word into a low dimensional feature, a recurrent neural network layer for learning caption sentence structure, and a multi-modal layer for combining visual and language information. In this model, the recurrent neural network layer is constructed by LSTM units, which are well known to be effective for learning and transferring sequence patterns. Moreover, this model has a unique structure in which the output of the convolution neural network layer is linked not only to the input of the initial state of the recurrent neural network layer but also to the input of the multimodal layer, in order to make use of visual information extracted from the image at each recurrent step for generating the corresponding textual caption. Through various comparative experiments using open data sets such as Flickr8k, Flickr30k, and MSCOCO, we demonstrated the proposed multimodal recurrent neural network model has high performance in terms of caption accuracy and model transfer effect.

• Bulletin of the Korean Chemical Society
• /
• v.31 no.5
• /
• pp.1143-1148
• /
• 2010

We investigated the separation of nitrogen heterocyclic compound (NHC) contained in a model coal tar fraction comprising four kinds of NHC [indole (In), quinoline (Q), iso-quinoline (iQ), quinaldine (Qu)], three kinds of bicyclic aromatic compound (BAC) [1-methylnaphthalene (1MN), 2-methylnaphthalene (2MN), dimethylnaphthalene (DMN) mixture with ten structural isomers (DMNs; regarded as one component)], biphenyl (Bp) and phenyl ether (Pe) by liquid membrane permeation (LMP). A batch-stirred tank was used as the permeation unit. An aqueous solution of saponin and n-hexane were used as the liquid membrane and the outer oil phase, respectively. Yield and selectivity of individual NHC was much larger than that of BAC, Bp and Pe. Increasing the initial mass fraction of the saponin to the membrane solution ($C_{sap,0}$) and the initial volume fraction of O/W emulsion to total liquid in a stirred tank (${\phi}_{OW,0}$) resulted in deteriorating the yield of individual NHC, but increasing the stirring speed (N) resulted in improving the yield of each NHC. With increasing $C_{sap,0}$, the selectivity of each NHC based on DMNs increased. Increasing ${\phi}_{OW,0}$ and N resulted in decreasing the selectivity of individual NHC based on DMNs. At an experimental condition fixed, the sequence of the yield and selectivity in reference to DMNs for each NHC was Q > Qu = iQ > In. Furthermore, we compared LPM method with methanol extraction method in view of the separation efficiency (yield, selectivity) of NHC.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.11 no.2
• /
• pp.261-267
• /
• 2011

One of powerful software paradigms for distributed systems is a mobile agent system. Since the usage of mobile agent systems is increased, a software development model to construct softwares efficiently for these systems is needed. Currently, UML is a widely used software development model. However, existing UML can not describe the necessary mobility of the mobile agent based software systems in explicit way. In this paper, the interaction diagrams of UML(sequence diagrams and communication diagrams) are extended and used to express the mobility of the mobile agents including the functions of platforms of mobile agent systems in three ways. For a case study, we applied the extended diagrams to a distributed file searching using mobile agents, and we confirmed these diagrams can describe the function and mobility of mobile agents very well.

• Journal of the Korean Society for Railway
• /
• v.20 no.1
• /
• pp.142-155
• /
• 2017

It is anticipated that, in cases in which different train operators share railway network, conflicts may occur among train operators with regard to train operation rate, train priorities, and arrival and departure time; besides this, during times when there is high demand for trains, operators will request train operations intensively, steadily increasing train conflict phenomena. In the present study, train operation sequence, minimum headway, arrival and departure time, train priorities, etc., were analyzed, and while using train departure times as decision variables in variably given train schedules, by adjusting train time requested by train operators, and finally rejecting the train times in cases in which conflict resolution is impossible, so that various constraints can be satisfied, a train slot allocation model was suggested to find the objective function, that is, the maximum number of train slots that can be practically applicable to railway operation.

• Korean Journal of Veterinary Research
• /
• v.50 no.4
• /
• pp.311-317
• /
• 2010

Recently the micropig has been developed as human disease model. The dental and orofacial region of micropig is similar to that of humans, so it has been used for testing implant materials and techniques. The purpose of this study is on dental image at each age using radiography and computed tomography. Total twenty-two male micropigs, two or three animals of each 1, 3, 5, 7, 9, 12, 18 and 24 months old, were given radiographic examinations. After general anesthesia, extra- and intra-oral radiographic technique and computed tomographic scans were performed to assess the dental characterization of micropigs. The total deciduous dental formula comprised 28 teeth and was depicted as Di 3/3, Dc 1/1, Dp 3/3. The total permanent dental formula comprised 44 teeth and was depicted as I 3/3, C 1/1, P 4/4, M 3/3. Hypodontia of the first premolars was common in the micropig. The permanent teeth erupted from 3 to 24 month after birth. The sequence of eruption of the permanent teeth was M1, P1, I3, C, M2, I1 + P3 + P4, P2, I2, M3. Dental imaging enables visualization of the unerupted teeth and gives more information about the development of the teeth. The growth pattern of the teeth obtained through radiographic and computed tomographic examination provides basic data in the micropig as animal model for dental research.

• The KIPS Transactions:PartB
• /
• v.17B no.6
• /
• pp.405-412
• /
• 2010

Previous background generation techniques showed bad performance in complex environments since they used only temporal contexts. To overcome this problem, in this paper, we propose a new background generation method which incorporates spatial as well as temporal contexts of the image. This enabled us to obtain 'clean' background image with no moving objects. In our proposed method, first we divided the sampled frame into m*n blocks in the video sequence and classified each block as either static or non-static. For blocks which are classified as non-static, we used MRF framework to model them in temporal and spatial contexts. MRF framework provides a convenient and consistent way of modeling context-dependent entities such as image pixels and correlated features. Experimental results show that our proposed method is more efficient than the traditional one.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.11 no.2
• /
• pp.426-434
• /
• 2010

This study is experimentally performed to evaluate the strength reduction behavior and its statistical properties of plain woven glass/epoxy composites. The results indicate that the major impact damage of plain woven glass/epoxy composites is the fiber breakage and matrix crack, whereas the dominant impact damage of unidirectional carbon/epoxy laminates is the delamination, which depends on the stacking sequence. The residual strength prediction models, previously proposed on unidirectional laminates, are applied to evaluate the residual strength of plain woven glass/epoxy composites with impact damage. Among these models, the results by Caprino and Avva's model have a good agreement with the experimental results. To investigate the variability of residual strength of the impacted composite materials, a statistical model was proposed and its results were in conformance with the experimental results regardless of their thickness.

• Economic and Environmental Geology
• /
• v.24 no.4
• /
• pp.409-419
• /
• 1991

A gravity survey was conducted in the western Marquette district, Michigan, to delineate the subsurface structure and the relationship of the Proterozoic Marquette Range Supergroup rocks (Precambrian X) and Archean basement (Precambrian W) where the Republic, Michigan River, and Marquette troughs join. In order to accomplish these purposes, three hundred and forty gravity stations were established in the area of $380km^2$. Positive anomalies are associated with the Precambria X, metasedimentary sequence which has a higer density with respect to the Precambrian W, basement rocks. The dominant positive gravity anomalies follow the axes of the three troughs which are filled with Precambrian X rocks. Subsurface structure was modelled by using the Talwani method. Gravity model studies indicate that the Marquette trough is asymetrically shaped and steeply dipping at the north edge except in the eastern part of the study area. The interpretive results obtained from two dimensional model studies suggest that the basement structure of the study area is relatively flat, and that the troughs were formed contemporaneously.