• Korean Journal of Remote Sensing
• /
• v.20 no.2
• /
• pp.139-152
• /
• 2004

This paper proposes to improve the results of image classification with spatial region growing segmentation by using an MRF-based classifier. The proposed approach is to re-classify the pixels in the boundary area, which have high probability of having classification error. The MRF-based classifier performs iteratively classification using the class parameters estimated from the region growing segmentation scheme. The proposed method has been evaluated using simulated data, and the experiment shows that it improve the classification results. But, conventional MRF-based techniques may yield incorrect results of classification for remotely-sensed images acquired over the ground area where has complicated types of land-use. A multistage MRF-based iterative class-modification in boundary is proposed to alleviate difficulty in classifying intricate land-cover. It has applied to remotely-sensed images collected on the Korean peninsula. The results show that the multistage scheme can produce a spatially smooth class-map with a more distinctive configuration of the classes and also preserve detailed features in the map.

• Korean Journal of Remote Sensing
• /
• v.34 no.2_1
• /
• pp.179-190
• /
• 2018

The purpose of this study is an analysis of the risk of soil erosion before and after the maintenance of riparian area using the Revised Universal Soil Loss Equation (RUSLE) model based on GIS and digitizing data. To analysis of soil erosion loss in the study area, land cover maps, topographical maps, soil maps, precipitation and other data were used. After digitizing the riparian area of the Geumhogang, the area is divided into administrative district units, respectively. Amount of soil loss was classified into 5 class according to the degree of loss. Totally, 1 and 5 class were decreased, and 2-4 class were increased. Daegu and Yeongcheon decreased the area of 5 class, and Gyeongsan did not have area of 5 class. The reason for this is thought to be the decrease of the 5 class area due to the park construction, expansion of artificial facilities, and reduction of agricultural land. Simplification of riverside for river dredging and park construction has increased the flow rate of the riverside and it is considered that the amount of soil erosion has increased.

• Journal of the Korea Society of Computer and Information
• /
• v.20 no.2
• /
• pp.21-28
• /
• 2015

In this paper, we propose a prototype-based classification learning by using the nearest-neighbor rule. The nearest-neighbor is applied to segment the class area of all the training data into spheres within which the data exist from the same class. Prototypes are the center of spheres and their radii are computed by the mid-point of the two distances to the farthest same class point and the nearest another class point. And we transform the prototype selection problem into a set covering problem in order to determine the smallest set of prototypes that include all the training data. The proposed prototype selection method is based on a greedy algorithm that is applicable to the training data per class. The complexity of the proposed method is not complicated and the possibility of its parallel implementation is high. The prototype-based classification learning takes up the set of prototypes and predicts the class of test data by the nearest neighbor rule. In experiments, the generalization performance of our prototype classifier is superior to those of the nearest neighbor, Bayes classifier, and another prototype classifier.

• Journal of Korea Multimedia Society
• /
• v.11 no.7
• /
• pp.1025-1034
• /
• 2008

In this paper, we developed GridGML for efficiently supplying a GML and visualizing the map with partitioning map and caching method to a mobile device. In order to overcome the weighting of a file, which is the biggest weakness of a GML, GridGML extracts only the most necessary parts for the visualization of the map among GML attributes, and makes the file light as a class instance by applying an offset value. GridGML manages a partition based on the visualization area of a mobile device to visualize the map to a mobile device in real time, and transmits the partition area by serializing it for the benefit of transmission. Also, the received partition area is compounded in a mobile device and is visualized by being partitioned again as four visible areas based on the display of a mobile device. Then, the area is managed by applying a caching algorithm in consideration of repetitiveness for a received map for the efficient operation of resources. Also, in order to prevent the delay in transmission time as regards the instance density area of the map, an adaptive map partition mechanism is proposed for maintaining the transmission time uniformly.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.20 no.3
• /
• pp.368-374
• /
• 2010

Image segmentation is the basic step in the field of the image processing for pattern recognition, environment recognition, and context analysis. The Otsu's automatic threshold selection, which determines the optimal threshold value to maximize the between class scatter using the distribution information of the normalized histogram of a image, is the famous method among the various image segmentation methods. For the automatic threshold selection proposed by Otsu, it is difficult to determine the optimal threshold value by considering the sub-region characteristic of the image because the Otsu's algorithm analyzes the global histogram of a image. In this paper, to alleviate this difficulty of Otsu's image segmentation algorithm and to improve image segmentation capability, the original image is divided into several sub-images by using context fuzzy c-means algorithm. The proposed fuzzy Otsu threshold algorithm is applied to the divided sub-images and the several threshold values are obtained.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.23 no.1
• /
• pp.65-70
• /
• 2023

In this paper, we used a deep learning model to detect and segment flame and smoke in real time from fires. To this end, well known U-NET was used to separate and divide the flame and smoke of the fire using multi-class. As a result of learning using the proposed technique, the values of loss error and accuracy are very good at 0.0486 and 0.97996, respectively. The IOU value used in object detection is also very good at 0.849. As a result of predicting fire images that were not used for learning using the learned model, the flame and smoke of fire are well detected and segmented, and smoke color were well distinguished. Proposed method can be used to build fire prediction and detection system.

• The Journal of the Korea Contents Association
• /
• v.21 no.3
• /
• pp.23-32
• /
• 2021

Semantic image segmentation technology in the field of computer vision is a technology that classifies an image by dividing it into pixels. This technique is also rapidly improving performance using a machine learning method, and a high possibility of utilizing information in units of pixels is drawing attention. However, this technology has been raised from the early days until recently for 'lack of detailed segmentation' problem. Since this problem was caused by increasing the size of the label map, it was expected that the label map could be improved by using the edge map of the original image with detailed edge information. Therefore, in this paper, we propose a post-processing algorithm that maintains semantic image segmentation based on learning, but modifies the resulting label map based on the edge map of the original image. After applying the algorithm to the existing method, when comparing similar applications before and after, approximately 1.74% pixels and 1.35% IoU (Intersection of Union) were applied, and when analyzing the results, the precise targeting fine segmentation function was improved.

• The Journal of the Acoustical Society of Korea
• /
• v.22 no.8
• /
• pp.727-739
• /
• 2003

In this paper, we propose a generation method of regression classes for adaptation in the HM-Net (Hidden Markov Network) system. The MLLR (Maximum Likelihood Linear Regression) adaptation approach is applied to the HM-Net speech recognition system for expressing the characteristics of speaker effectively and the use of HM-Net in various tasks. For the state level sharing, the context domain state splitting of PDT-SSS (Phonetic Decision Tree-based Successive State Splitting) algorithm, which has the contextual and time domain clustering, is adopted. In each state of contextual domain, the desired phoneme classes are determined by splitting the context information (classes) including target speaker's speech data. The number of adaptation parameters, such as means and variances, is autonomously controlled by contextual domain state splitting of PDT-SSS, depending on the context information and the amount of adaptation utterances from a new speaker. The experiments are performed to verify the effectiveness of the proposed method on the KLE (The center for Korean Language Engineering) 452 data and YNU (Yeungnam Dniv) 200 data. The experimental results show that the accuracies of phone, word, and sentence recognition system increased by 34∼37%, 9%, and 20%, respectively, Compared with performance according to the length of adaptation utterances, the performance are also significantly improved even in short adaptation utterances. Therefore, we can argue that the proposed regression class method is well applied to HM-Net speech recognition system employing MLLR speaker adaptation.

• Journal of IKEEE
• /
• v.25 no.2
• /
• pp.273-279
• /
• 2021

In this paper, we propose an algorithm that shortens the learning time by performing individual learning using partitioning the deep learning structure. The proposed algorithm consists of four processes: network classification origin setting process, feature vector extraction process, feature noise removal process, and class classification process. First, in the process of setting the network classification starting point, the division starting point of the network structure for effective feature vector extraction is set. Second, in the feature vector extraction process, feature vectors are extracted without additional learning using the weights previously learned. Third, in the feature noise removal process, the extracted feature vector is received and the output value of each class is learned to remove noise from the data. Fourth, in the class classification process, the noise-removed feature vector is input to the multi-layer perceptron structure, and the result is output and learned. To evaluate the performance of the proposed algorithm, we experimented with the Extended Yale B face database. As a result of the experiment, in the case of the time required for one-time learning, the proposed algorithm reduced 40.7% based on the existing algorithm. In addition, the number of learning up to the target recognition rate was shortened compared with the existing algorithm. Through the experimental results, it was confirmed that the one-time learning time and the total learning time were reduced and improved over the existing algorithm.

• The Magazine of the IEIE
• /
• v.30 no.5
• /
• pp.95-95
• /
• 2003

현재의 라우팅 프로토콜은 다양한 사용자 요구를 만족시켜주기 위해서는 네트워크의 처리량을 최대화하고 동시에 사용자의 요구 시 QoS를 보장해주는 기법이 요구되고 있다. 기존의 최단경로 라우팅 프로토콜은 단일경로 라우팅으로 인해 병목현상의 단점을 지니고 있다. 즉, 원천과 목적지간 최단경로는 낮은 활용도를 나타내는 경로들이 많이 존재하지만 단일경로를 선택하므로서 폭주(congestion)의 발생확률이 높다. 최근에 들어 사용자의 QoS 요구 시, 다양한 QoS를 패킷 네트워크에서 처리할 수 있도록 IETF에서 DiffServ, RSVP, MPLS 등과 같은 패킷 QoS 기법에 대한 표준화 작업이 진행중이며, 그 중에서 Diffserv 네트워크가 대표적이다. 따라서 본 논문에서는 이 DiffServ 네트워크상에서 다양하게 유입되는 트래픽의 종류에 따라 사용자의 응용에 적절히 대응하여 트래픽을 처리하는 라우팅 기법 및 알고리즘을 연구하고 기존의 최선형 (Best effort) 트래픽을 처리하기 위한 트래픽 분산 라우팅 프로토콜 (Traffic-Balanced Rout-ing Protocol''TBRP)을 제안하였으며, 최적의 중간 노드를 선택하여 높은 순위의 상호형 데이터를 처리하기 위한 계층적 라우팅 프로토콜(또ierarchicalTra(fic-Scheduling Routing Protocol : HTSRP)을 연구하였다. 본 연구에서 제시한 프로토콜은 유, 무선망의 통합에 따른 다양한 엑세스망과 백본망에 유연한 트래픽 처리기법으로서 계층적 라우팅 알고리즘으로 적합하였다. 본 실험에서는 사용자의 QoS요청 시 제공되는 상호형 또는 스트리 밍 데이터를 위한 HTSRP_Q(Hierarchical Traffic-Scheduling Routing Pro-tocol for QoS)에 대해 성능이 우수함을 입증하였으며, 각 엑세스 단에서 요청하는 QoS 파라미터에 따라 자원을 최적화하여 QoS를 보장하고, 특히 지연에 민감한 트래픽을 처리하였으며, 제안한 프로토콜을 이용하여 사용자 요구 트래픽 종류에 따라 대화형 클래스, 스트리밍 클래스, 높은 순위의 상호형 클래스, 낮은 순위의 상호형 클래스, 그리고 background 클래스등 5개의 서비스 클래스로 분리하여 트래픽 특성에 맞게 처리할 수 있었다. QoS 관련 실험에서는 QoS 요청데이터를 균등하게 1에서 10Mbps 사이에 분포하였고 연결된 호에 대한 지속시간은 5분으로 하였다. 이러한 환경에서 프로토콜을 MaRS에 의해 실험을 하였고 기존의 거리-벡터 라우팅과 링크-상태 라우팅 프로토콜과 비교해서 처리량, 메시지 손실, 블럭킹율 등에서 비교적 우위의 성능을 확인할 수 있었으며, 특히, 차별화된 서비스의 특성에 맞게 라우팅 기법을 적용하므로서 망의 효율성과 안정성을 꾀할 수가 있었다. 연결 수 대 처리량에서는 HTSRP 프로토콜이 연결이 적을 때 DVR, LSR보다 우월하였으며 특히, 선형을 유지하였다. 연결 수 대 패킷 손실에서 HTSRP프로토콜에서 메시지 손실은 연결의 수가 낮거나 높을 때 다른 DVR과 LSR 라우팅 프로토콜과 유사한 결과를 나타내었다. Hotspo에서 TBRP, HTSRP프로토콜은 hotspot 연결의 수가 9일 때까지 DVR, LSR 보다 좋은 처리량를 나타냈고 HTSRP는 연결의 수가 6 이상일 때 가장 높은 처리량을 나타내었다. 일반 트래픽과 QoS 트래픽이 흔재할 경우는 트래픽이 증가할수록 HTSRP_Q가 가장 월등하였으며 , 로드가 증가할수록 낮은 블록킹률을 나타내었다. 본 논문에서는 점대점 전송을 기반으로 하였다. 앞으로 다양한 응용 S/W는 멀티캐스트 기반이 예상되므로 멀티캐스트 라우팅에 대한 연구가 필요하다. 본 논문의 프로토콜은 원천과 목적지간의 최단경로가 폭주상태가 아닌 해당 중간 노드를 이용한다. 최단경로의 모든 링크상의 트래픽 부하가 낮을 때 중간노드의 사용은 지연을 증가시킨다. 향후 최적의 성능을 위해 보완이 필요하다. 아울러, 2계위에서는 일반 트래픽과 QoS 트래픽이 혼재할 때 자동으로 네트워크의 효율적을 고려한 방법 선택이 필요하다.