• 전자공학회논문지
• /
• 제53권7호
• /
• pp.49-55
• /
• 2016

디지털 포렌식 영상은 여러 가지 영상타입으로 위 변조되어 유통되는 심각한 문제가 대두되어 있다. 이러한 문제를 해결하기 위하여, 본 논문에서는 포렌식 영상의 분류 알고리즘을 제안한다. 제안된 알고리즘은 여러 가지 영상타입의 그레이 레벨 co-occurrence 행렬의 특성 중에서 콘트라스트와 에너지 그리고 영상의 엔트로피로 21-dim.의 특징벡터를 추출하고, 결정나무 플랜에서 분류학습을 위하여 PPCA를 이용하여 2-dim.으로 차원을 축소한다. 포렌식 영상의 분류 테스트는 영상 타입들의 전수조합에서 수행되었다. 실험을 통하여, TP (True Positive)와 FN (False Negative)을 검출하고, 제안된 알고리즘의 성능평가에서 민감도 (Sensitivity)와 1-특이도 (1-Specificity)의 AUROC (Area Under Receiver Operating Characteristic) 커브 면적은 0.9980으로 'Excellent(A)' 등급임을 확인하였다. 산출된 최소평균 판정에러 0.0179에서 분류할 포렌식 영상타입이 모두 포함되어 분류 효율성이 높다.

• 한국콘텐츠학회논문지
• /
• 제11권12호
• /
• pp.497-503
• /
• 2011

본 연구에서는 환경음 인식 성능의 향상을 위하여 GMM의 훈련 방식에 MCE 도입을 제안하였다. 이는 환경음 데이터 모델링에 사용할 분류오류함수를 정의할 때 해당 클래스의 로그우도 뿐 아니라 다른 클래스의 로그우도도 같이 고려함으로써 변별력 있는 분류가 이뤄질 수 있게 한다. 모델의 파라미터는 전체 클래스를 고려한 손실함수를 정의하고, GPD(generalized probabilistic descent)알고리즘을 이용하여 추정하였다. 제안된 방법의 인식 성능 비교를 위해 모두 9가지 환경음을 전처리 과정과 MFCC(mel-frequency cepstral coefficients)를 이용하여 12차 특징을 추출하고, 이를 혼합 성분의 수에 따라 GMM 분류 실험을 행하였다. 실험 결과에 따르면 혼합 성분을 19개 사용한 경우에서 MCE 훈련 방식이 평균 87.06%의 인식률로 가장 좋은 성능을 보였다. 이 결과로 제안한 MCE 훈련 방식이 환경음 인식에서 GMM의 훈련 방식으로 효과적으로 사용될 수 있음을 확인하였다.

• Journal of Trauma and Injury
• /
• 제22권1호
• /
• pp.108-115
• /
• 2009

Purpose: We designed a retrospective study to measure the accuracy of the ICD-10 (International Classification of Disease-10) code for trauma patients. We also analyzed the error of the ICISS (International Classification of Disease based Injury Severity Score) due to a missing or an incorrect ICD-10 code. Methods: For the measuring the accuracy of the ICD-10 code for trauma patients in a tertiary teaching hospital's emergency department, two board certified emergency physician performed a retrospective chart review. The ICD-10 code was classified as a main code or a sub-code. The main code was defined as the code of the main department of treatment, and the sub-code was defined as a code other than the main code. We calculated and compared two ICISS for each patient one by using both the existing code and the other by using a corrected code. We compared the proportions of severe trauma (defined as an ICISS less than 0.9) between when the existing code and the corrected code was used respectively. Results: We reviewed the records of 4287 trauma patients who had been treated from July 2008 to November 2008. The accuracy of the main code, the sub-code of emergency department, main-code, the sub-code of hospitalized patients were 97.1%, 59.8%, 98.2% and 57.0%, respectively. Total accuracy of the main and sub-code of emergency department and of hospitalized patients were 91.4% and 58.6%. The number of severe trauma patients increased from 33 to 49 when the corrected code was used in emergency department and increased from 35 to 60 in hospitalized patients. Conclusion: The accuracy of the sub-code was lower than that of the main code. A missing or incorrect subcode could cause an error in the ICISS and in the number of patients with severe trauma.

• 한국운동역학회지
• /
• 제16권4호
• /
• pp.95-103
• /
• 2006

The purpose of this study was to review the relevant literature about coaching and thereupon, survey the coaching methods used for golf lesson to reinterpret them and thereby, describe in view of kinetics the swing errors committed frequently by amateur golfers and suggest more scientific golf coaching methods. For this purpose, kinetic elements were divided into accuracy and power ones and therewith, the variables affecting such elements were identified. For this study, a total of 60 amateur golfer were sampled, and their swing forms were photographed with two high-speed digital cameras, and the resultant images were analyzed to determine the errors of each form kinetically, which would be analyzed again with the program V1-5000. The kinetic elements could be identified as accuracy, power and accuracy & power. Thus, setup and trajectory were classified into accuracy elements, while differences of inter-joint angles, cocking and delayed hitting. Lastly, timing and axial movement were classified into accuracy & power elements. Three errors were identified in association with setup. The errors related with trajectory elements accounted for most (6) of the 20 errors. Three errors were determined for inter-joint angle differences, and one error was associated with cocking and delayed hitting. Lastly, one error was classified into timing error, while five errors were associated with axial movement. Finally, as a result of arranging the errors into a cross table, it was found that the errors were associated with each other between take-back and back-swing, take-back and follow-through, back-swing and back-swing top, and between back-swing and down-swing. Namely, an error would lead to other error repeatedly. So, it is more effective to identify all the errors for every form and correct them comprehensively rather than single out the errors and correct them one by one.

• 한국융합학회논문지
• /
• 제12권12호
• /
• pp.65-74
• /
• 2021

최근 기계 번역 기술과 자동 노이즈 생성 방법론을 기반으로 한국어 맞춤법 교정 연구가 활발히 이루어지고 있다. 해당 방법론들은 노이즈를 생성하여 학습 셋과 데이터 셋으로 사용한다. 이는 학습에 사용된 노이즈 외의 노이즈가 테스트 셋에 포함될 가능성이 낮아 정확한 성능 측정이 어렵다는 한계점이 존재한다. 또한 실제적인 오류 유형 분류 기준이 없어 연구마다 사용하는 오류 유형이 다르므로 질적 분석에 어려움을 겪고 있다. 이를 해결하기 위해 본 논문은 딥러닝 기반 한국어 맞춤법 교정 연구를 위한 새로운 '오류 유형 분류 체계'를 제안하며 이를 바탕으로 기존 상용화 한국어 맞춤법 교정기(시스템 A, 시스템 B, 시스템 C)에 대한 오류 분석을 수행하였다. 분석결과, 세 가지 교정 시스템들이 띄어쓰기 오류 외에 본 논문에서 제시한 다른 오류 유형은 교정을 잘 수행하지 못했으며 어순 오류나 시제 오류의 경우 오류 인식을 거의 하지 못함을 알 수 있었다.

• 대한원격탐사학회지
• /
• 제30권2호
• /
• pp.219-226
• /
• 2014

분광혼합분석 결과로 얻어지는 각 물체의 점유비율을 활용하면 보다 세밀한 분류가 가능하다. 이는 복잡한 도심지역의 피복분류 뿐만 아니라 혼효림이 많은 한반도 임상분류에 적합한 분류기법이 될 수 있다. 효과적인 임상분류를 위해서는 무엇보다 적절한 endmember의 추출이 선행되어야 하는데, 기존에 주로 사용되었던 기하학적 방법(geometric endmember selection)은 분광특성이 유사한 산림지역에 적합하지 않다. 본 연구에서는 영상에서 직접 순수한 화소를 추출하는 기법 중의 하나인 IEA(Iterative Error Analysis)와 침엽수와 활엽수의 분광특성을 이용하여 실험지역을 대표할 수 있는 각각의 endmember를 자동으로 추출하였다. CASI(Compact Airborne Spectrographic Imager) 영상의 두 지역에 대하여 분광혼합분석을 이용한 분류를 수행한 결과, 분류 정확도는 각각 86%와 90%로, 제안한 기법이 실험대상지역을 대표하는 침엽수와 활엽수의 endmember를 적절하게 추출한 것으로 나타났다. 분광혼합분석 기법을 이용한 보다 효과적인 분류를 위해서 분류항목 외 기타물질을 endmember로 고려하는 연구가 필요할 것으로 보인다.

• International Journal of Advanced Culture Technology
• /
• 제4권2호
• /
• pp.38-42
• /
• 2016

In the growing data saturated world, the question of "whether data can be used" has shifted to "can it be utilized effectively?" More data is being generated and utilized than ever before. As the collection of data increases, data mining techniques also must become more and more accurate. Thus, to ensure this data is effectively utilized, the analysis of the data must be efficient. Interpretation of results from the analysis of the data set presented, have their own on the basis it is possible to obtain the desired data. In the data mining method a decision tree, clustering, there is such a relationship has not yet been fully developed algorithm actually still impact of various factors. In this experiment, the classification method of data mining techniques is used with easy decision tree. Also, it is used special technology of one R and J48 classification technique in the decision tree. After selecting a rule that a small error in the "one rule" in one R classification, to create one of the rules of the prediction data, it is simple and accurate classification algorithm. To create a rule for the prediction, we make up a frequency table of each prediction of the goal. This is then displayed by creating rules with one R, state-of-the-art, classification algorithm while creating a simple rule to be interpreted by the researcher. While the following can be correctly classified the pattern specified in the classification J48, using the concept of a simple decision tree information theory for configuring information theory. To compare the one R algorithm, it can be analyzed error rate and accuracy. One R and J48 are generally frequently used two classifications${\ldots}$

• 농업과학연구
• /
• 제47권2호
• /
• pp.205-217
• /
• 2020

In this study, a deep learning-based tillage boundary detection method for autonomous tillage by a tractor was developed, which consisted of image cropping, object classification, area segmentation, and boundary detection methods. Full HD (1920 × 1080) images were obtained using a RGB camera installed on the hood of a tractor and were cropped to 112 × 112 size images to generate a dataset for training the classification model. The classification model was constructed based on convolutional neural networks, and the path boundary was detected using a probability map, which was generated by the integration of softmax outputs. The results show that the F1-score of the classification was approximately 0.91, and it had a similar performance as the deep learning-based classification task in the agriculture field. The path boundary was determined with edge detection and the Hough transform, and it was compared to the actual path boundary. The average lateral error was approximately 11.4 cm, and the average angle error was approximately 8.9°. The proposed technique can perform as well as other approaches; however, it only needs low cost memory to execute the process unlike other deep learning-based approaches. It is possible that an autonomous farm robot can be easily developed with this proposed technique using a simple hardware configuration.

• 대한원격탐사학회지
• /
• 제34권1호
• /
• pp.141-150
• /
• 2018

Crop type classification is essential for supporting agricultural decisions and resource monitoring. Remote sensing techniques, especially using hyperspectral imagery, have been effective in agricultural applications. Hyperspectral imagery acquires contiguous and narrow spectral bands in a wide range. However, large dimensionality results in unreliable estimates of classifiers and high computational burdens. Therefore, reducing the dimensionality of hyperspectral imagery is necessary. In this study, the Random Forest (RF) classifier was utilized for dimensionality reduction as well as classification purpose. RF is an ensemble-learning algorithm created based on the Classification and Regression Tree (CART), which has gained attention due to its high classification accuracy and fast processing speed. The RF performance for crop classification with airborne hyperspectral imagery was assessed. The study area was the cultivated area in Chogye-myeon, Habcheon-gun, Gyeongsangnam-do, South Korea, where the main crops are garlic, onion, and wheat. Parameter optimization was conducted to maximize the classification accuracy. Then, the dimensionality reduction was conducted based on RF variable importance. The result shows that using the selected bands presents an excellent classification accuracy without using whole datasets. Moreover, a majority of selected bands are concentrated on visible (VIS) region, especially region related to chlorophyll content. Therefore, it can be inferred that the phenological status after the mature stage influences red-edge spectral reflectance.

• 한국측량학회지
• /
• 제35권1호
• /
• pp.1-10
• /
• 2017

An UAV (Unmanned Aerial Vehicle) is a flight system that is designed to conduct missions without a pilot. Compared to traditional airborne-based photogrammetry, UAV-based photogrammetry is inexpensive and can obtain high-spatial resolution data quickly. In this study, we aimed to classify the land cover using high-spatial resolution images obtained using a UAV. An RGB camera was used to obtain high-spatial resolution orthoimage. For accurate classification, multispectral image about same areas were obtained using a multispectral sensor. A DSM (Digital Surface Model) and a modified NDVI (Normalized Difference Vegetation Index) were generated using images obtained using the RGB camera and multispectral sensor. Pixel-based classification was performed for twelve classes by using the RF (Random Forest) method. The classification accuracy was evaluated based on the error matrix, and it was confirmed that the proposed method effectively classified the area compared to supervised classification using only the RGB image.