• ETRI Journal
• /
• v.29 no.6
• /
• pp.736-744
• /
• 2007

We propose a fast macroblock (MB) mode prediction and decision algorithm based on temporal correlation for P-slices in the H.264/AVC video standard. There are eight block types for temporal decorrelation, including SKIP mode based on rate-distortion (RD) optimization. This scheme gives rise to exhaustive computations (search) in the coding procedure. To overcome this problem, a thresholding method for fast inter mode decision using a MB tracking scheme to find the most correlated block and RD cost of the correlated block is suggested for early stop of the inter mode determination. We propose a two-step inter mode candidate selection method using statistical analysis. In the first step, a mode is selected based on the mode information of the co-located MB from the previous frame. Then, an adaptive thresholding scheme is applied using the RD cost of the most correlated MB. Secondly, additional candidate modes are considered to determine the best mode of the initial candidate modes that does not satisfy the designed thresholding rule. Comparative analysis shows that a speed-up factor of up to 70.59% is obtained when compared with the full mode search method with a negligible bit increment and a minimal loss of image quality.

• Proceedings of the KIEE Conference
• /
• 1989.07a
• /
• pp.241-245
• /
• 1989

A knowledge based expert system is a computer program that emulates the reasoning process of a human expert in a specific problem domain. This paper presents an expert system to diagnose the various faults in power system. The developed expert system is represented considering two points; the possibility of solution and the fast processing speed. As uncertainties exist in the facts and rules which comprise the knowledge base of the expert system, Certainty Factor, which is based on the confirmation theory is used for the inexact reasoning. Also, as the diagnosis problem requires the inductive reasoning process in nature, the solution is imperfect and not unique in general. So the expert system is designed to generate all the possible hypothesis in order of the possibility and also it can explain the propagation procedure of the faults for each solution using the built in backtracking mechanism. In realization of the expert system, the processing speed is greatly dependent upon the problem representation, reasoning scheme and search strategy. So, in this paper the fault diagnosis problem itself is analysed from the view point of Artificial Intelligence and as a result, the expert system has the following basic features. 1) The certainty factor is adopted in the inference engine for inexact reasoning. 2) Problem apace is represented using the problem reduction technique. 3) Bidirectional reasoning scheme is used. 4) Best first search strategy is adopted for rapid processing. The expert system was developed us ing PROLOG language.

• The Journal of the Korea Contents Association
• /
• v.7 no.12
• /
• pp.311-321
• /
• 2007

For fast and robust face detection, this paper proposes an approach for face detection using fusion of heterogeneous template matching. First, we detect skin regions using a model of skin color which covers various illumination and races. After reducing a search space by region labelling and filtering, we apply template matching with skin color and edge to the detected regions. Finally, we detect a face by finding the best choice of template fusion. Experimental results show the proposed approach is more robust in skin color-like environments than with a single template matching and is fast by reducing a search space to face candidate regions. Also, using a global accumulator can reduce excessive space requirements of template matching.

• Korean Medical Education Review
• /
• v.10 no.1
• /
• pp.35-42
• /
• 2008

Purpose : Since PBL was first developed by Howard Barrows at McMaster, it has been adopted as one of the best teaching and learning methods in medical schools throughout the world. However, the educational superiority of PBL relative to traditional approaches is less clear. Given the somewhat extensive resources required for the operation of PBL curriculum, this gives reason for concern. The aim of this study is to review experiences of PBL in other medical schools and learn how to implement PBL in our school. Methods : This study was undertaken in two stages. In the first stage, PBL curricular examples in 7 medical schools (University of Pennsylvania, University of Melbourne, University of Maastricht, McMaster University, Flinders University, Harvard medical school. University of California at L.A.) were collected and summarized. In the second stage, a careful search for articles of journals published since 2000 regarding PBL group assessment, effectiveness of PBL and group facilitation skills was conducted. Results : PBL is generally introduced in a core curriculum in undergraduate medical education. Relating to small group assessment, the perception of students has been well developed. but the current PBL assessment tool needs to be revised, to develop thinking skills of students. The PBL graduates considered themselves as having much better interpersonal skills, better competencies in problem solving and self-directed learning than the non-PBL graduates. Tutors used various techniques to raise awareness, facilitate the group process and direct learning. Conclusions : The following three aspects can be regarded as important in this study. First, to implement PBL in our school more effectively, it might be considered, which curriculum content can be best learned with PBL. Second, to enhance students' thinking skills during PBL, a new assessment tool needs to be developed. Third, tutors' competencies are important to facilitate, group process, so it would be worthwhile including in staff development.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.7
• /
• pp.3393-3412
• /
• 2017

As a main challenge in network virtualization, virtual network embedding problem is increasingly important and heuristic algorithms are of great interest. Aiming at the problems of poor correlation in node embedding and link embedding, long distance between adjacent virtual nodes and imbalance resource consumption of network components during embedding, we herein propose a two-stage virtual network embedding algorithm NA-PVNM. In node embedding stage, resource requirement and breadth first search algorithm are introduced to sort virtual nodes, and a node fitness function is developed to find the best substrate node. In link embedding stage, a path fitness function is developed to find the best path in which available bandwidth, CPU and path length are considered. Simulation results showed that the proposed algorithm could shorten link embedding distance, increase the acceptance ratio and revenue to cost ratio compared to previously reported algorithms. We also analyzed the impact of position constraint and substrate network attribute on algorithm performance, as well as the utilization of the substrate network resources during embedding via simulation. The results showed that, under the constraint of substrate resource distribution and virtual network requests, the critical factor of improving success ratio is to reduce resource consumption during embedding.

• Korean Journal of Remote Sensing
• /
• v.33 no.4
• /
• pp.423-436
• /
• 2017

This paper presents a deep learning-based road segmentation framework from very high-resolution orthophotos. The proposed method uses Deep Convolutional Autoencoders for end-to-end mapping of orthophotos to road segmentations. In addition, a set of post-processing steps were applied to make the model outputs GIS-ready data that could be useful for various applications. The optimization of the model's parameters is explained which was conducted via grid search method. The model was trained and implemented in Keras, a high-level deep learning framework run on top of Tensorflow. The results show that the proposed model with the best-obtained hyperparameters could segment road objects from orthophotos at an average accuracy of 88.5%. The results of optimization revealed that the best optimization algorithm and activation function for the studied task are Stochastic Gradient Descent (SGD) and Exponential Linear Unit (ELU), respectively. In addition, the best numbers of convolutional filters were found to be 8 for the first and second layers and 128 for the third and fourth layers of the proposed network architecture. Moreover, the analysis on the time complexity of the model showed that the model could be trained in 4 hours and 50 minutes on 1024 high-resolution images of size $106{\times}106pixels$, and segment road objects from similar size and resolution images in around 14 minutes. The results show that the deep learning models such as Convolutional Autoencoders could be a best alternative to traditional machine learning models for road segmentation from aerial photographs.

• Proceedings of the Korea Institute of Fire Science and Engineering Conference
• /
• 2008.11a
• /
• pp.304-309
• /
• 2008

본 논문은 기존 기발된 재실자의 탈출 계획 및 예측 프로그램 개발에 대해 길 찾기를 위한 알고리즘으로는 Dijkstra 알고리즘, Best-First Search 알고리즘, Johnson 알고리즘 등이 있으며 가장 안정적으로 알고리즘 구현이 가능한 A*알고리즘을 적용하였다. 따라서, 본 개발 프로그램은 재실자가 대피 목적지를 향한 최적의 길 찾기를 이용하여 가장 가까운 거리에 있는 탈출구를 효율적으로 찾을 수가 있으며 재실자의 사실감 있는 대피 이동 동선의 구현을 위해 기존의 경직된 경로를 매끄럽게 구현하였다. 탈출구는 흐름율과 정체 반경을 적용하여 재실자가 탈출구에 밀집하였을 경우 병목 현상이 발생하도록 하여 대피현상이 실제 상황과 유사하도록 프로그램을 구축하였다. 본 대피프로그램은 실제 건물의 CAD도면을 import 가능하도록 구축함으로서 대피평가시간을 절약할 수 있도록 하였다.

• Proceedings of the KSRS Conference
• /
• 2005.10a
• /
• pp.501-504
• /
• 2005

In general, most incomplete communication link setup between satellite and ground station right after separation from launcher come from less accurate orbital vector ground station uses to track the satellite because only predicted orbital state vector is available during first few orbits. This paper describes the developed procedure for successful initial acquisition for KOMPSAT-2 using scanning functions ofK13 antenna system with predicted orbital information. Azimuth scan, raster scan, spiral scan functions were tested with KOMPSA Tl under intentionally degraded orbital information for antenna operation. Through tests, spiral scan function was decided to be best search scan among 3 scans. Developed procedure can assure the successful acquisition only if azimuth offset and time offset value are within +/-2deg and +/-30sec, respectively.

• Journal of Advanced Marine Engineering and Technology
• /
• v.21 no.3
• /
• pp.293-301
• /
• 1997

Edge detection is the first step and very important step in image analysis. We cast edge detec¬tion as a problem in cost maximization. This is acheived by the formulation of a cost function that evaluates the quality of edge configurations. The cost function can be used as a basis for compar¬ing the performances of different detectors. We used a Genetic Algorithm for maximizing cost func¬tion. Genetic algorithms are a class of adaptive search techniques that have been intensively stud¬ied in recent years and have been prone to converge prematurely before the best solution has been found. This paper shows that carefully chosen modifications(three factors of the crossover opera¬tor) are implemented can be effective in alleviating this problem.

• Steel and Composite Structures
• /
• v.16 no.6
• /
• pp.681-701
• /
• 2014

The main aim of this study is to propose an efficient meta-heuristic algorithm for topology optimization of geometrically nonlinear single layer domes by serially integration of computational advantages of firefly algorithm (FA) and particle swarm optimization (PSO). During the optimization process, the optimum number of rings, the optimum height of crown and tubular section of the member groups are determined considering geometric nonlinear behaviour of the domes. In the proposed algorithm, termed as FA-PSO, in the first stage an optimization process is accomplished using FA to explore the design space then, in the second stage, a local search is performed using PSO around the best solution found by FA. The optimum designs obtained by the proposed algorithm are compared with those reported in the literature and it is demonstrated that the FA-PSO converges to better solutions spending less computational cost emphasizing on the efficiency of the proposed algorithm.