• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.36 no.5
• /
• pp.423-432
• /
• 2018

Surface water mapping has been widely used in various remote sensing applications. Water indices have been commonly used to distinguish water bodies from land; however, determining the optimal threshold and discriminating water bodies from similar objects such as shadows and snow is difficult. Deep learning algorithms have greatly advanced image segmentation and classification. In particular, FCN (Fully Convolutional Network) is state-of-the-art in per-pixel image segmentation and are used in most benchmarks such as PASCAL VOC2012 and Microsoft COCO (Common Objects in Context). However, these data sets are designed for daily scenarios and a few studies have conducted on applications of FCN using large scale remotely sensed data set. This paper aims to fine-tune the pre-trained FCN network using the CRMS (Coastwide Reference Monitoring System) data set for surface water mapping. The CRMS provides color infrared aerial photos and ground truth maps for the monitoring and restoration of wetlands in Louisiana, USA. To effectively learn the characteristics of surface water, we used pre-trained the DeepWaterMap network, which classifies water, land, snow, ice, clouds, and shadows using Landsat satellite images. Furthermore, the DeepWaterMap network was fine-tuned for the CRMS data set using two classes: water and land. The fine-tuned network finally classifies surface water without any additional learning process. The experimental results show that the proposed method enables high-quality surface mapping from CRMS data set and show the suitability of pre-trained FCN networks using remote sensing data for surface water mapping.

• Journal of Korea Multimedia Society
• /
• v.13 no.5
• /
• pp.651-662
• /
• 2010

This paper describes a method to identify objects for autonomous navigation of an outdoor mobile robot. To identify objects, the robot recognizes the object from an image taken by moving robot on outdoor environment. As a beginning, this paper presents the candidates for a segment of region to building of artificial object, sky and trees of natural objects. Then we define their characteristics individually. In the process, we segment the regions of the objects included by preprocessing using multiple features. Multiple features are HSI, line segments, context information, hue co-occurrence matrix, principal components and vanishing point. An analysis of building identifies the geometrical properties of building facet such as wall region, windows and entrance. The building as intersection in vertical and horizontal line segment of vanishing point extracts the mesh. The wall region of building detect by merging the mesh of the neighbor parallelograms that have similar colors. The property estimates the number of story and rooms in the same floors by merging skewed parallelograms of the same color. We accomplish the result of image segmentation using multiple features and the geometrical properties analysis of object through experiments.

• Journal of the HCI Society of Korea
• /
• v.6 no.1
• /
• pp.9-16
• /
• 2011

The previous researches of 3D object modeling have been performed in a limited environment where a target object only exists. However, in order to model an object in the real environment, we need to consider a dynamic environment, which has various objects and a frequently changing background. Therefore, this paper presents a segmentation and tracking method for a foreground which includes a target object in the dynamic environment. By using depth information than color information, the foreground region can be segmented and tracked more robustly. In addition, the foreground region can be tracked on the sequential images by referring depth distributions of the foreground region because both the position and the status in the consecutive images of the foreground region are almost unchanged. Experimental results show that our proposed method can robustly segment and track the foreground region in various conditions of the real environment. Moreover, as an application of the proposed method, it is presented a method for modeling an object extracting the object regions from the foreground region that is segmented and tracked.

• Journal of Korea Multimedia Society
• /
• v.9 no.2
• /
• pp.160-171
• /
• 2006

In this paper, we propose a face detection method using region segmentation to deal with complex images that have various environmental changes such as mixed background and light changes. To reduce the detection error rate due to background elements of the images, we segment the images with the JSEG method. We choose candidate regions of face based on the ratio of skin pixels from the segmented regions. From the candidate regions we detect face regions by using location and color information of eyes and eyebrows. In the experiment, the proposed method works well with the images that have several faces and different face size as well as mixed background and light changes.

• Journal of The Korean Society of Agricultural Engineers
• /
• v.64 no.1
• /
• pp.15-26
• /
• 2022

3D open-field farm model developed from UAV (Unmanned Aerial Vehicle) data could make crop monitoring easier, also could be an important dataset for various fields like remote sensing or precision agriculture. It is essential to separate crops from the non-crop area because labeling in a manual way is extremely laborious and not appropriate for continuous monitoring. We, therefore, made a 3D open-field farm model based on UAV images and developed a crop segmentation model using a supervised machine learning algorithm. We compared performances from various models using different data features like color or geographic coordinates, and two supervised learning algorithms which are SVM (Support Vector Machine) and KNN (K-Nearest Neighbors). The best approach was trained with 2-dimensional data, ExGR (Excess of Green minus Excess of Red) and z coordinate value, using KNN algorithm, whose accuracy, precision, recall, F1 score was 97.85, 96.51, 88.54, 92.35% respectively. Also, we compared our model performance with similar previous work. Our approach showed slightly better accuracy, and it detected the actual crop better than the previous approach, while it also classified actual non-crop points (e.g. weeds) as crops.

• Journal of the Institute of Electronics Engineers of Korea SC
• /
• v.43 no.5 s.311
• /
• pp.89-100
• /
• 2006

The carotid intima-media thickness (IMT) is very important, because the severity of it is an independent predictor of transient cerebral ischemia, stroke, and coronary events such as myocardial infarction. The conventional image processing to measure the IMT has not been satisfactory, because the methods have relied on the manual section drawing and a regional segmentation by differential estimation. We propose a new image processing technology effective to extract features from the carotid artery image whose pixels have the directional vector properties with composed color distribution. The technique we presented here is not by differential variation but by verification of the layer properties of carotid artery image. Iterated vertical and horizontal analysis and segmentation of the IMT image show the vector characteristics. This new technique makes it possible to cluster the layers statistically, and to classify mathematical correlation between regions and resulting in correct measurements of thickness and its variation. The advantages and effectiveness of this approach are applicable to region process and character extraction of such a vector image.

• Proceedings of the Korean Information Science Society Conference
• /
• 1999.10b
• /
• pp.533-535
• /
• 1999

자궁경부진 세포인식 시스템에 있어서 가장 중요한 것이 영상처리를 이용하여 세포핵과 세포질을 추출하여 세포의 형태적인 정보를 알아내는 과정이다. 기존의 전역 thresholding 기법이나 region growing의 경우는 pap smear 검사를 통해 얻어진 세포 영상을 분할할 수 있는 region growing 기법을 제안한다. 제안된 region growing 기법은 초기에 seed를 검출할 때 local threshold growing 기법을 제안한다. 제안된 region growing 기법은 초기에 seed를 검출할 때 local threshold 개념을 도입하여 seed의 검출을 고르게 하고, 2가지 확장 조건을 사용하여 영역을 확장한다. 첫 번째 확장 조건은 비정상 세포나 artifact가 많아서 어둡게 나타나는 영상이나 세포질과 배경의 경계가 뚜렷하지 않아서 세포질의 구별이 어려운 영상의 영역 분할이 가능하도록 그 특성을 반영하고, 두 번째 조건은 세포가 흡수하는 빛의 양이 일정하다는 가정으로 영상에서의 지역 특성(gray level, color 등을 반영한다. 제안된 기법은 정상세포 영상뿐만 아니라 비정상 세포 영상에 대하여 over-segment나 under-segment하는 경우를 줄여서 영역 분할에 좋은 결과를 보인다.

• Proceedings of the Korean Information Science Society Conference
• /
• 1999.10b
• /
• pp.422-424
• /
• 1999

본 논문에서는 질감 특징과 신경회로망을 이용한 유방암조직영상의 분할 방법을 제안한다. 신경회로망의 입력 노드에 사용될 질감 특징을 얻기 위해 10개의 영상에 대해 각 영역(양성세포핵, 음성세포핵, 배경)에서 10개씩의 화소를 선택하고, 그 화소를 중심으로 하는 5$\times$5 영역 30개를 획득, 총 300개의 영역에 대해 R, G, B 각각의 밴드에서 18개의 질감특징을 추출한다. 54개의 입력노드, 28개의 은닉노드, 3개의 출력노드의 구조를 가진 신경회로망을 구성하고, 역전파 학습 알고리즘을 사용하여 신경회로망을 최대오차율이 10-3보다 작을 때까지 학습시킨다. 학습에 의해 획득되어진 분류기를 이용하여 유방암 조직 세포영상을 양성세포핵, 음성세포핵, 배경부분으로 자동 분할한다.

• 한국HCI학회:학술대회논문집
• /
• 2006.02a
• /
• pp.786-791
• /
• 2006

색상 전이는 스타일 전이, 색이 바랜 사진의 복원, 색상화, 색상의 보정에 사용될 수 있는 기법이다. 본 연구에서는 기존 색상 전이의 문제점을 해결하기 위해서 영상 분할 기반의 색상전이 기법을 제시한다. 영상에서 색상의 가장 의미있는 최소 단위를 픽셀로 보고 있는 기존 연구에 반해서, 본 연구에서는 영상 조각을 영상에서 가장 의미 있는 최소 단위로 보고 색상 전이를 수행한다. 영상 분할 기반의 색상 전이를 통해서 기존 연구에서 발생할 수 있었던 픽셀간의 코헤런스 문제를 해결한다. 또한 영상 분할 기반으로 했을 때에 생길 수 있는 경계 문제를 해결하기 위한 새로운 방법을 제시한다. 제시된 기법을 이용해서 색상 전이의 응용인 스타일 전이에 적용한다.

• Proceedings of the IEEK Conference
• /
• 2009.05a
• /
• pp.34-36
• /
• 2009

This paper proposes an effective lane detection algorithm using hugh transform and lane model. The proposed lane detection algorithm includes two major components, i.e., lane marks segmentation and an exact lane extraction using a novel postprocessing technique. The first step is to segment lane marks from background images using HSV color model. Then, a novel postprocessing is used to detect an exact lane using Hugh transform and lane models(linear and curved lane models). The postprocessing consists of three parts, i.e, thinning process, Hugh Transform and filtering process. We divide input image into three regions of interests(ROIs). Based on lane curve function(LCF), we can detect an exact lane from various extracted lane lines. The lane models(linear and curved lane mode]) are used in order to judge whether each lane segment is fit or not in each ROIs. Experimental results show that the proposed scheme is very effective in lane detection.