Support Vector Machine Classification Using Training Sets of Small Mixed Pixels: An Appropriateness Assessment of IKONOS Imagery

Byeong Hyeok Yu,Kwang Hoon Chi 大韓遠隔探査學會 2008 大韓遠隔探査學會誌 Vol.24 No.5

Many studies have generally used a large number of pure pixels as an approach to training set design. The training set are used, however, varies between classifiers. In the recent research, it was reported that small mixed pixels between classes are actually more useful than larger pure pixels of each class in Support Vector Machine (SVM) classification. We evaluated a usability of small mixed pixels as a training set for the classification of high-resolution satellite imagery. We presented an advanced approach to obtain a mixed pixel readily, and evaluated the appropriateness with the land cover classification from IKONOS satellite imagery. The results showed that the accuracy of the classification based on small mixed pixels is nearly identical to the accuracy of the classification based on large pure pixels. However, it also showed a limitation that small mixed pixels used may provide insufficient information to separate the classes. Small mixed pixels of the class border region provide cost-effective training sets, but its use with other pixels must be considered in use of high-resolution satellite imagery or relatively complex land cover situations.

Application of Multiple Threshold Values for Accuracy Improvement of an Automated Binary Change Detection Model

( Byeong Hyeok Yu ),( Kwang Hoon Chi ) 대한원격탐사학회 2009 大韓遠隔探査學會誌 Vol.25 No.3

Multi-temporal satellite imagery can be changed into a transform image that emphasizes the changed area only through the application of various change detection techniques. From the transform image, an automated change detection model calculates the optimal threshold value for classifying the changed and unchanged areas. However, the model can cause undesirable results when the histogram of the transform image is unbalanced. This is because the model uses a single threshold value in which the sign is either positive or negative and its value is constant (e.g. -1, 1), regardless of the imbalance between changed pixels. This paper proposes an advanced method that can improve accuracy by applying separate threshold values according to the increased or decreased range of the changed pixels. It applies multiple threshold values based on the cumulative producer`s and user`s accuracies in the automated binary change detection model, and the analyst can automatically extract more accurate optimal threshold values. Multi-temporal IKONOS satellite imagery for the Daejeon area was used to test the proposed method. A total of 16 transformation results were applied to the two study sites, and optimal threshold values were determined using accuracy assessment curves. The experiment showed that the accuracy of most transform images is improved by applying multiple threshold values. The proposed method is expected to be used in various study fields, such as detection of illegal urban building, detection of the damaged area in a disaster, etc.

Application of Multiple Threshold Values for Accuracy Improvement of an Automated Binary Change Detection Model

Yu, Byeong-Hyeok,Chi, Kwang-Hoon The Korean Society of Remote Sensing 2009 大韓遠隔探査學會誌 Vol.25 No.3

Multi-temporal satellite imagery can be changed into a transform image that emphasizes the changed area only through the application of various change detection techniques. From the transform image, an automated change detection model calculates the optimal threshold value for classifying the changed and unchanged areas. However, the model can cause undesirable results when the histogram of the transform image is unbalanced. This is because the model uses a single threshold value in which the sign is either positive or negative and its value is constant (e.g. -1, 1), regardless of the imbalance between changed pixels. This paper proposes an advanced method that can improve accuracy by applying separate threshold values according to the increased or decreased range of the changed pixels. It applies multiple threshold values based on the cumulative producer's and user's accuracies in the automated binary change detection model, and the analyst can automatically extract more accurate optimal threshold values. Multi-temporal IKONOS satellite imagery for the Daejeon area was used to test the proposed method. A total of 16 transformation results were applied to the two study sites, and optimal threshold values were determined using accuracy assessment curves. The experiment showed that the accuracy of most transform images is improved by applying multiple threshold values. The proposed method is expected to be used in various study fields, such as detection of illegal urban building, detection of the damaged area in a disaster, etc.

Deep-Learning-based Automatic Identification of Wildlife Species : A Case Study in Sobaeksan National Park, Korea

Byeong-Hyeok Yu 강원대학교 산림과학연구소 2018 강원대학교 산림과학연구소 학술대회 Vol.2018 No.09

Camera traps are mainly used to detect wildlife in protected areas. The captured images are interpreted by the human eye. Such visual interpretation is not only time consuming, but also makes it difficult to maintain data consistency when investigators changed. Recently, deep learning has been detecting object identification, counts, and image description in imagery with high accuracy. In this paper, we introduce the camera trap data processor that can automatically database wildlife species identification by deep learning. The Sobaeksan National Park's Jukryong eco-corridor was selected as a study area. Through the image-tracking algorithm, the minimum bounding rectangle of the wild animal object was detected and each frame was used as a training image. For deep learning, we used a convolutional neural network (CNN) technique, which is preferred in image recognition field. Open source libraries (OpenCV, TensorFlow, and Keras) were used to implement the model, and the software was developed through Python. The study results showed possibilities that it can reduce the survey time and minimize human errors.

Support Vector Machine Classification Using Training Sets of Small Mixed Pixels: An Appropriateness Assessment of IKONOS Imagery

Yu, Byeong-Hyeok,Chi, Kwang-Hoon The Korean Society of Remote Sensing 2008 大韓遠隔探査學會誌 Vol.24 No.5

Many studies have generally used a large number of pure pixels as an approach to training set design. The training set are used, however, varies between classifiers. In the recent research, it was reported that small mixed pixels between classes are actually more useful than larger pure pixels of each class in Support Vector Machine (SVM) classification. We evaluated a usability of small mixed pixels as a training set for the classification of high-resolution satellite imagery. We presented an advanced approach to obtain a mixed pixel readily, and evaluated the appropriateness with the land cover classification from IKONOS satellite imagery. The results showed that the accuracy of the classification based on small mixed pixels is nearly identical to the accuracy of the classification based on large pure pixels. However, it also showed a limitation that small mixed pixels used may provide insufficient information to separate the classes. Small mixed pixels of the class border region provide cost-effective training sets, but its use with other pixels must be considered in use of high-resolution satellite imagery or relatively complex land cover situations.

고점도 물질의 슬롯 다이 운전 조건에 관한 실험적 연구

유병혁(Byeong Hyeok Yu),유상석(Sangseok Yu),박기호(Ki Ho Park),오상현(Sang Hyun Oh),전원표(Won Pyo Chun),김성일(Sung Il Kim) 대한기계학회 2019 대한기계학회 춘추학술대회 Vol.2019 No.4

INFORMATION - 재난대응을 위한 오픈소스 GIS(FOSS4G) 활용사례

유병혁,Yu, Byeong-Hyeok 한국방재협회 2018 방재저널 Vol.20 No.2

유발된 굴절이상이 보행패턴에 미치는 영향

최재혁(Jae Hyeok Choi),문병연(Byeong-Yeon Moon),유동식(Dong-Sik Yu),조현국(Hyun Gug Cho),김상엽(Sang-Yeob Kim) 한국안광학회 2018 한국안광학회지 Vol.23 No.3

Purpose: To investigate the effects of induced refractive errors on gait patterns. Methods: Three-two subjects of average age 22.50±2.22 years were participated in this study. To induce binocular myopia and hyperopia, spherical lenses of ±0.50 D, ±1.00 D, ±2.00 D, ±3.00 D, ±4.00 D, and ±5.00 D were used. Gait patterns (step length and cadence) were measured on a treadmill at a speed 4 km/h, upper-body sway was also evaluated using motion sensor during gait. After each repeated measurements in refractive errors, the measurement values measured in the full corrected condition were compared with those measured. Results: The step length was significantly shortened and cadence was significantly increased from myopia of –1.00 D compared to the full corrected condition. The area of upper-body sway was extended to the anterior-posterior and left-right position compared with the full correction condition during gait, while myopic and hyperopic refractive errors were induced. Conclusions: Uncorrected myopic refraction error above –1.00 D was a factor for changing the gait pattern during gait.

유발된 굴절이상에 따른 보행 패턴의 변화와 시각 기능과의 상관성 분석

최재혁(Jae Hyeok Choi),조현국(Hyun Gug Cho),문병연(Byeong-Yeon Moon),유동식(Dong-Sik Yu),김상엽(Sang-Yeob Kim) 한국안광학회 2020 한국안광학회지 Vol.25 No.3

Purpose: To investigate changes in the gait pattern caused by induced refractive errors and analyze the correlation with visual function factors. Methods: We enrolled 40 healthy subjects with an average age of 22.23±1.99 years. To induce binocular myopia and hyperopia, S ±1.00, ±2.00, and ±3.00 D were used. The gait pattern depending on each induced refractive error was measured on a treadmill at a speed of 4 km/h using OptoGait. After each repeated test at each level of refractive error, the values measured in the full corrected condition were compared to those measured. In addition, the correlation between the gait pattern and visual function elements was analyzed. Results: The step length shortened and cadence increased significantly from induced myopia by +1.00 D and induced hyperopia by -1.00 D compared to the full corrected condition. In hyperopia, the visual function factor that is most relevant to gait pattern changes was analyzed as the reduced accommodative amplitude. Conclusions: Regardless of the refractive error type, the uncorrected refractive error was a visual state that temporarily interfered with the optimal gait pattern.

Adjuvant Cytokine-Induced Killer Cell Immunotherapy for Hepatocellular Carcinoma: A Real-World Experience from Two Large-Volume Centers in Korea

( Jun Sik Yoon ),( Byeong Geun Song ),( Jeong-hoon Lee ),( Hyo Young Lee ),( Sun Woong Kim ),( Young Chang ),( Eun Ju Cho ),( Su Jong Yu ),( Dong Hyun Sinn ),( Yoon Jun Kim ),( Joon Hyeok Lee ),( Jung 대한간학회 2018 춘·추계 학술대회 (KASL) Vol.2018 No.1

Aims: Several randomized controlled trials have shown that adjuvant immunotherapy with autologous cytokine-induced killer (CIK) cells prolongs recurrence-free survival (RFS) after curative treatment for hepatocellular carcinoma (HCC). We investigated the efficacy of adjuvant immunotherapy with activated CIK cells in real-world clinical practice. Methods: A total of 370 patients with stage I or II HCC who underwent curative surgical resection or radiofrequency ablation at Seoul National University Hospital or Samsung Seoul Medical Center were included in this study. Among them, 71 patients received CIK cell immunotherapy after curative treatment and 299 patients did not. Propensity score matching with a 1:1 ratio was conducted to avoid possible bias and 64 pairs of matched patients were generated. The primary endpoint was RFS and secondary endpoints included overall survival. Results: After a propensity score matching, there was no statistical difference in variables including treatment modalities, HCC stage, presence of cirrhosis, alpha-fetoprotein level between the two groups. The median follow-up duration was 16.0 months (interquartile range, 7.8 - 27.0). The immunotherapy group did not reach median RFS and the control group showed 33.5 months of median RFS (P=0.001 by log-rank test). CIK cell adjuvant therapy reduced the risk of tumor recurrence or death by 65% (hazard ratio [HR], 0.35; 95% confidence interval [CI], 0.18-0.68, P=0.002, Figure 1). No patient (0%) in the immunotherapy group and 5 patients (7.8%) in the control group died during the study period. However, the difference in OS between two groups failed to reach statistical significance (P=0.060 by Firth’s method, Figure 2). Multivariable Cox regression analysis showed that immunotherapy was an independent predictor for HCC recurrence (adjusted HR, 0.16; 95% CI, 0.04-0.65; P=0.010). Conclusions: The adjuvant immunotherapy with autologous CIK cells prolongs RFS in patients with HCC after curative therapy in a real-world setting.