• 대한원격탐사학회지
• /
• 제40권1호
• /
• pp.57-69
• /
• 2024

Crop classification plays a vitalrole in monitoring agricultural landscapes and enhancing food production. In this study, we explore the effectiveness of Long Short-Term Memory (LSTM) models for crop classification, focusing on distinguishing between apple and rice crops. The aim wasto overcome the challenges associatedwith finding phenology-based classification thresholds by utilizing LSTM to capture the entire Normalized Difference Vegetation Index (NDVI)trend. Our methodology involvestraining the LSTM model using a reference site and applying it to three separate three test sites. Firstly, we generated 25 NDVI imagesfrom the Sentinel-2A data. Aftersegmenting study areas, we calculated the mean NDVI values for each segment. For the reference area, employed a training approach utilizing the NDVI trend line. This trend line served as the basis for training our crop classification model. Following the training phase, we applied the trained model to three separate test sites. The results demonstrated a high overall accuracy of 0.92 and a kappa coefficient of 0.85 for the reference site. The overall accuracies for the test sites were also favorable, ranging from 0.88 to 0.92, indicating successful classification outcomes. We also found that certain phenological metrics can be less effective in crop classification therefore limitations of relying solely on phenological map thresholds and emphasizes the challenges in detecting phenology in real-time, particularly in the early stages of crops. Our study demonstrates the potential of LSTM models in crop classification tasks, showcasing their ability to capture temporal dependencies and analyze timeseriesremote sensing data.While limitations exist in capturing specific phenological events, the integration of alternative approaches holds promise for enhancing classification accuracy. By leveraging advanced techniques and considering the specific challenges of agricultural landscapes, we can continue to refine crop classification models and support agricultural management practices.

• 대한원격탐사학회:학술대회논문집
• /
• 대한원격탐사학회 2006년도 Proceedings of ISRS 2006 PORSEC Volume I
• /
• pp.297-300
• /
• 2006

The knowledge of fuel load and composition is important for planning and managing the fire hazard and risk. However, fuel mapping is extremely difficult because fuel properties vary at spatial scales, change depending on the seasonal situations and are affected by the surrounding environment. Remote sensing has potential of reduction the uncertainty in mapping fuels and offers the best approach for improving our abilities. This paper compared the results of object-oriented classification to a pixel-based classification for fuel type map derived from Hyperion hyperspectral data that could be enable to provide this information and allow a differentiation of material due to their typical spectra. Our methodological approach for fuel type map is characterized by the result of the spectral mixture analysis (SMA) that can used to model the spectral variability in multi- or hyperspectral images and to relate the results to the physical abundance of surface constitutes represented by the spectral endmembers. Object-oriented approach was based on segment based endmember selection, while pixel-based method used standard SMA. To validate and compare, we used true-color high resolution orthoimagery

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제12권6호
• /
• pp.2748-2760
• /
• 2018

This paper proposes an audio event classification method based on convolutional neural networks (CNNs). CNN has great advantages of distinguishing complex shapes of image. Proposed system uses the features of audio sound as an input image of CNN. Mel scale filter bank features are extracted from each frame, then the features are concatenated over 40 consecutive frames and as a result, the concatenated frames are regarded as an input image. The output layer of CNN generates probabilities of audio event (e.g. dogs bark, siren, forest). The event probabilities for all images in an audio segment are accumulated, then the audio event having the highest accumulated probability is determined to be the classification result. This proposed method classified thirty audio events with the accuracy of 81.5% for the UrbanSound8K, BBC Sound FX, DCASE2016, and FREESOUND dataset.

• 디지털산업정보학회논문지
• /
• 제14권4호
• /
• pp.69-77
• /
• 2018

The malignant melanoma accounts for about 1 to 3% of the total malignant tumor in the West, especially in the US, it is a disease that causes more than 9,000 deaths each year. Generally, skin lesions are difficult to detect the features through photography. In this paper, we propose a computer-aided diagnosis algorithm based on deep learning for classification of malignant melanoma and benign skin tumor in RGB channel skin images. The proposed deep learning model configures the tumor lesion segmentation model and a classification model of malignant melanoma. First, U-Net was used to segment a skin lesion area in the dermoscopic image. We could implement algorithms to classify malignant melanoma and benign tumor using skin lesion image and results of expert's labeling in ResNet. The U-Net model obtained a dice similarity coefficient of 83.45% compared with results of expert's labeling. The classification accuracy of malignant melanoma obtained the 83.06%. As the result, it is expected that the proposed artificial intelligence algorithm will utilize as a computer-aided diagnosis algorithm and help to detect malignant melanoma at an early stage.

• 정보처리학회논문지D
• /
• 제16D권1호
• /
• pp.11-26
• /
• 2009

심혈관계 질환의 진단 위해서 복합 진단 지표를 이용한 출현 패턴 기반의 분류 기법을 제안하였다. 복합 진단 지표 적용을 위해서 심박동변이도의 선형/비선형적 특징들을 세 가지 누운 자세에 대해 분석하였고 ST-segments로부터 4개의 진단 지표를 추출하였다. 이 논문에서는 질환진단을 위해서 필수 출현 패턴을 이용한 분류 모델을 제안하였다. 이 분류 기법은 환자 그룹의 질환 패턴들을 발견하며, 이러한 출현 패턴은 심혈관계 질환 환자들에서는 빈발하지만 정상인 그룹에서는 빈발하지 않는 패턴들이다. 제안된 분류 알고리즘의 평가를 위해서 120명의 협심증(AP: angina pectrois) 환자, 13명의 급성관상동맥증후군(ACS: acute coronary syndrome) 환자 그리고 128명의 정상인 데이터를 사용하였다. 실험 결과 복합 지표를 사용하였을 때, 세 그룹의 분류에 대한 정확도는 약 88.3%였다.

• 한국산림과학회지
• /
• 제98권3호
• /
• pp.324-330
• /
• 2009

본 연구의 목적은 KOMPSAT-2 영상을 이용하여 대상지 내 주요 수종의 이산화탄소 흡수량을 추정하는 것이다. 우선, KOMPSAT-2 영상을 분할영상분류방법으로 산림내 수종을 임상단위로 분류하였다. 분류된 임상의 평균 수관직경을 추정한 후, 이를 각 수종의 면적에 대비시켜 수종별 본수를 추정하였다. 평균 흉고직경은 지상조사를 통해 유도된 수관직경과 흉고직경의 관계식을 이용하여 추정하였다. 이와 같이 추정된 수종별 흉고직경과 본수를 기존의 바이오매스 추정식에 대입하여 수종별 바이오매스를 추정하였다. 바이오매스에 '기후변화에 관한 정부간 패널(IPCC)'의 가이드라인에 따른 계수를 적용하여 임상단위의 이산화탄소 흡수량을 추정했다. 본 연구의 임상단위 수종별 이산화탄소 흡수량 추정 접근방식은 향후, 표본점 단위의 이산화탄소 흡수량을 임상단위 이산화탄소 흡수량으로 확장시키는데 활용될 수 있다.

• 전자공학회논문지
• /
• 제50권12호
• /
• pp.224-231
• /
• 2013

이 논문에서는 비정형 혼합 오디오 신호에서 청취자에게 전달 되도록 의도된 주된 신호의 종류를 검출해 낼 수 있는 방법을 제안한다. 주된 신호의 종류는 음성, 음악, 음향효과로 정하였으며, 인텐서티 기반의 발생/소멸 패턴에서 추출할 수 있는 특징을 사용하여 그들을 구별할 수 있는 방법을 소개한다. 청취자가 주어진 오디오 신호에서 주된 신호를 받아들이는 주관적인 평가를 반영하기 위해서, 웹기반의 평가시스템을 도입하여 18시간의 다양한 종류의 장르 비디오의 오디오를 평가하였다. 실험을 통하여 비디오의 장르별로 각기 다른 성능을 보이지만 가능성 있는 (음성위주의 토크쇼의 경우 86.7%, 액션 영화 49.3%)정확도를 보였다.

• 한국GIS학회:학술대회논문집
• /
• 한국GIS학회 2003년도 공동 춘계학술대회 논문집
• /
• pp.588-595
• /
• 2003

High resolution panchromatic satellite images collected by sensors such as IRS-1C/D and KOMPSAT-1 have a spatial resolution of approximately 6 ${\times}$ 6 ㎡, making them very attractive for urban applications. However, the spectral information present in these images is very limited. In order to overcome this limitation, an object-oriented classification approach is used to identify basic land cover types in urban areas. Before an image can be classified it is segmented at different aggregation levels using a multiresolution segmentation approach. In the course of this segmentation various statistical as well as topological information is collected for each segment. Based on this information it is possible to classify image objects and to arrive at much better results than by looking only at single pixels. Using an image recorded by KOMPSAT-1 over the City of Vienna a land cover classification was carried out for two areas. One was used to set up the rules for the different land cover types. The second subset was classified based on these rules, only adjusting some of the functions governing the classification process.

• Journal of Trauma and Injury
• /
• 제35권1호
• /
• pp.12-18
• /
• 2022

Purpose: Pedicle screw fixation provides 3-column stabilization, multidimensional control, and a higher rate of interbody fusion. Although computed tomography (CT) is recommended for the postoperative assessment of pedicle screw fixation, its use is limited due to the radiation exposure dose. The purpose of this preliminary retrospective study was to assess the clinical usefulness of low-dose mobile cone-beam CT (CBCT) for the postoperative evaluation of pedicle screw fixation. Methods: The author retrospectively reviewed postoperative mobile CBCT images of 15 patients who underwent posterior pedicle screw fixation for spinal disease from November 2019 to April 2020. Pedicle screw placement was assessed for breaches of the bony structures. The breaches were graded based on the Heary classification. Results: The patients included 11 men and four women, and their mean age was 66±12 years. Of the 122 pedicle screws, 34 (27.9%) were inserted in the thoracic segment (from T7 to T12), 82 (67.2%) in the lumbar segment (from L1 to L5), and six (4.9%) in the first sacral segment. Although there were metal-related artifacts, the image of the screw position (according to Heary classification) after surgery could be assessed using mobile CBCT at all levels (T7-S1). Conclusions: Mobile CBCT was accurate in determining the location and integrity of the pedicle screw and identifying the surrounding bony structures. In the postoperative setting, mobile CBCT can be used as a primary modality for assessing the accuracy of pedicle screw fixation and detecting postoperative complications.

• 전자공학회논문지B
• /
• 제33B권3호
• /
• pp.79-87
• /
• 1996

A new method for recognizing and locating partially occluded or overlapped two-dimensional objects regardless of their size, translation, and rotation, is presented. Dominant points approximating occuluding contoures of objects are generated by finding local maxima of smoothed k-cosine function, and then used to guide the contour segment matching procedure. Primitives between the dominant points are produced by projecting the local contours onto the line between the dominant points. Robust classification of primitives. Which is crucial for reliable partial shape matching, is performed using adaptive resonance theory (ART2). The matched primitives having similar scale factors and rotation angles are detected in the hough space to identify the presence of the given model in the object scene. Finally the translation vector is estimated by minimizing the mean squred error of the matched contur segment pairs. This model-based matching algorithm may be used in diveerse factory automation applications since models can be added or changed simply by training ART2 adaptively without modifying the matching algorithm.