• Journal of the Korea Society of Computer and Information
• /
• v.17 no.4
• /
• pp.67-74
• /
• 2012

In this paper, we propose an efficient dynamic background modelling method by using eigenbackground to extract moving objects from video stream. Even if a background model has been created, the model has to be updated to adapt to change due to several reasons such as weather or lighting. In this paper, we update a background model based on R-SVD method. At this time we define a change ratio of images and update the model dynamically according this value. Also eigenbackground need to be modelled by using sufficient training images for accurate models but we reorganize input images to reduce the number of images for training models. Through simulation, we show that the proposed method improves the performance against traditional eigenbackground method without background updating and a previous method.

• Journal of the Korea Society for Simulation
• /
• v.19 no.4
• /
• pp.199-208
• /
• 2010

Positions of tactical objects are represented as Time, Space and Position Information(TSPI) in modeling and simulations(M&S). The format and required information record for TSPI is investigated by referring the TSPI object model of the Test and Training Enabling Architecture(TENA), which has been developed by the United States Department of Defense. The most sophisticated tactical data link, Link-16 has a Precise Participant Location and Information (PPLI) message. We study the data format for exchanging TSPI data based on the PPLI message. To evaluate and track positions of tactical objects, we consider the Kalman filter for linear systems, and the extended Kalman filter and the unscented Kalman filter for nonlinear systems. Based on motion equations of a ballistic missile, the tracking performance for the trajectory of the ballistic missile is simulated by the unscented Kalman filter.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.3 no.1
• /
• pp.47-52
• /
• 2010

In this paper, we introduce a new approach for real-time 3D face pose discrimination based on active IR illumination from a monocular view of the camera. Under the IR illumination, the pupils appear bright. We develop algorithms for efficient and robust detection and tracking pupils in real time. Based on the geometric distortions of pupils under different face orientations, an eigen eye feature space is built based on training data that captures the relationship between 3D face orientation and the geometric features of the pupils. The 3D face pose for an input query image is subsequently classified using the eigen eye feature space. From the experiment, we obtained the range of results of discrimination from the subjects which close to the camera are from 94,67%, minimum from 100%, maximum.

• The Transactions of the Korea Information Processing Society
• /
• v.3 no.5
• /
• pp.1037-1045
• /
• 1996

In this paper, the parallel model for stereo camera is employed to find the heralding angle and the distance between a leading vehicle and the following vehicle, BART(Binocular Autonomous Research Team vehicle). Two TDNNs (Time Delay Neural Network) such as S-TDNN and A-TDNN are introduced to control BART. S-TDNN controls the speed of the following vehicle while A-TDNN controls the steering angle of BATR. A human drives BART to collect data which are used for training the said neural networks. The trained networks performed the vehicle tracking function satisfactorily under the same driving conditions performed by the human driver. The neural network approach has good portability which decreases costs and saves development time for the different types of vehicles.

• International Journal of Precision Engineering and Manufacturing
• /
• v.7 no.2
• /
• pp.18-24
• /
• 2006

This paper proposes a LuGre Model-Based Neural Network (MBNN) friction compensation algorithm for a linear motor stage. For matching the friction phenomena in both the motion-start region and the motion-reverse region, the LuGre dynamic model is employed into the proposed compensation algorithm. After training of the model-based neural network is completed, the estimated friction for compensation is obtained. From the obtained result we find that the new structure gains advantage over the non-friction compensation system on the performance of the compensator in both regions. The proposed compensator is evaluated and compared experimentally with an uncompensated system on a microcomputer controlled linear motor tracking system in the final section of the paper. The experimental results show the improvement on the maximum velocity error and the root mean square tracking error in the motion-start region ranges from 34% to 53% and from 53% to 75% respectively, and in the motion-reverse region from 48% to 65% and from 79% to 90% respectively.

• Proceedings of the Korean Information Science Society Conference
• /
• 1998.10c
• /
• pp.384-386
• /
• 1998

The use of hand gesture provides an attractive alternative to cumbersome interface devices for human-computer interaction(HCI). Many methods hand gesture recognition using visual analysis have been proposed such as syntactical analysis, neural network(NN), Hidden Markov Model(HMM) and so on. In our research, a HMMs is proposed for alphabetical hand gesture recognition. In the preprocessing stage, the proposed approach consists of three different procedures for hand localization, hand tracking and gesture spotting. The hand location procedure detects the candidated regions on the basis of skin-color and motion in an image by using a color histogram matching and time-varying edge difference techniques. The hand tracking algorithm finds the centroid of a moving hand region, connect those centroids, and thus, produces a trajectory. The spotting a feature database, the proposed approach use the mesh feature code for codebook of HMM. In our experiments, 1300 alphabetical and 1300 untrained gestures are used for training and testing, respectively. Those experimental results demonstrate that the proposed approach yields a higher and satisfying recognition rate for the images with different sizes, shapes and skew angles.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.8 no.3
• /
• pp.727-732
• /
• 2004

In this paper, we introduce a new approach for real-time 3D face pose discrimination based on active IR illumination from a monocular view of the camera. Under the IR illumination, the pupils appear bright. We develop algorithms for efficient and robust detection and tracking pupils in real time. Based on the geometric distortions of pupils under different face orientations, an eigen eye feature space is built based on training data that captures the relationship between 3D face orientation and the geometric features of the pupils. The 3D face pose for an input query image is subsequently classified using the eigen eye feature space. From the experiment, we obtained the range of results of discrimination from the subjects which close to the camera are from 94,67％, minimum from 100％, maximum.

• International Journal of Advanced Culture Technology
• /
• v.6 no.4
• /
• pp.52-57
• /
• 2018

The purpose of this paper is to develop simulations that can be used for virtual education in dentistry. The virtual education to be developed will be developed with clinical training and actual case data of tooth extraction. This development goal is to allow dental students to learn the necessary surgical techniques at the point of their choice, not going into the operating room, away from time, space, and physical limits. I want to develop content using VR. Oculus Rift HMD, Optical Based Outside-in Tracking System, Oculus Touch Motion Controller, and Headset as Input / Output Device. In this configuration, the optimization method is applied convergent, and when the operation of the VR contents is performed, the content data is extracted from the interaction analysis formed in the VR engine, and the data is processed by the content algorithm. It also computes events and dental operations generated within the 3D engine programming and generates corresponding events through data processing according to the input signal. The visualization information is output to the HMD using the rendering information. In addition, the operating room environment was constructed by studying lighting and material for actual operating room environment. We applied the ratio of actual space to virtual space and the ratio between character and actual person to create a spatial composition at a similar rate to actual space.

• Journal of Ocean Engineering and Technology
• /
• v.37 no.1
• /
• pp.38-48
• /
• 2023

To maintain the existing systems of ships and introduce autonomous operation technology, it is necessary to improve situational awareness through the sensor fusion of the automatic identification system (AIS) and automatic radar plotting aid (ARPA), which are installed sensors. This study proposes an algorithm for determining whether AIS and ARPA signals are sent to the same ship in real time. To minimize the number of errors caused by the time series and abnormal phenomena of heterogeneous signals, a tracking method based on the combination of the unscented Kalman filter and probabilistic data association filter is performed on ARPA radar signals, and a position prediction method is applied to AIS signals. Especially, the proposed algorithm determines whether the signal is for the same vessel by comparing motion-related components among data of heterogeneous signals to which the corresponding method is applied. Finally, a measurement test is conducted on a training ship. In this process, the proposed algorithm is validated using the AIS and ARPA signal data received by the voyage data recorder for the same ship. In addition, the proposed algorithm is verified by comparing the test results with those obtained from raw data. Therefore, it is recommended to use a sensor fusion algorithm that considers the characteristics of sensors to improve the situational awareness accuracy of existing ship systems.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2022.11a
• /
• pp.308-310
• /
• 2022

With the rise of the Al wave, task-oriented dialogue systems have become one of the popular research directions in academia and industry. Currently, task-oriented dialogue systems mainly adopt pipelined form, which mainly includes natural language understanding, dialogue state decision making, dialogue state tracking and natural language generation. However, pipelining is prone to error propagation, so many task-oriented dialogue systems in the market are only for single-round dialogues. Usually single- domain dialogues have relatively accurate semantic understanding, while they tend to perform poorly on multi-domain, multi-round dialogue datasets. To solve these issues, we developed a paddle-based multi-domain task-oriented Chinese dialogue system. It is based on NEZHA-base pre-training model and CrossWOZ dataset, and uses intention recognition module, dichotomous slot recognition module and NER recognition module to do DST and generate replies based on rules. Experiments show that the dialogue system not only makes good use of the context, but also effectively addresses long-term dependencies. In our approach, the DST of dialogue tracking state is improved, and our DST can identify multiple slotted key-value pairs involved in the discourse, which eliminates the need for manual tagging and thus greatly saves manpower.