• Journal of Cadastre & Land InformatiX
• /
• v.48 no.2
• /
• pp.213-225
• /
• 2018

For constructing three-dimensional (3D) spatial information occlusion region problem arises in the process of taking the texture of the building. In order to solve this problem, it is necessary to investigate the automation method to automatically recognize the occlusion region, issue it, and automatically complement the texture. In fact there are occasions when it is possible to generate a very large number of structures and occlusion, so alternatives to overcome are being considered. In this study, we attempt to apply an approach to automatically create an occlusion region based on learning by patterning the blocked region using the recently emerging deep learning algorithm. Experiment to see the performance automatic detection of people, banners, vehicles, and traffic lights that cause occlusion in building walls using two advanced algorithms of Convolutional Neural Network (CNN) technique, Faster Region-based Convolutional Neural Network (R-CNN) and Mask R-CNN. And the results of the automatic detection by learning the banners in the pre-learned model of the Mask R-CNN method were found to be excellent.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.8
• /
• pp.1976-1995
• /
• 2023

The anchor-free object detector CenterNet regards the object as a center point and predicts it based on the Gaussian circle region. For each object's center point, CenterNet directly regresses the width and height of the objects and finally gets the boundary range of the objects. However, the critical range of the object's center point can not be accurately limited by using the Gaussian circle region to constrain the prediction region, resulting in many low-quality centers' predicted values. In addition, because of the large difference between the width and height of different objects, directly regressing the width and height will make the model difficult to converge and lose the intrinsic relationship between them, thereby reducing the stability and consistency of accuracy. For these problems, we proposed a center point prediction method based on the Gaussian elliptic region and a size component regression method based on the small solution space. First, we constructed a Gaussian ellipse region that can accurately predict the object's center point. Second, we recode the width and height of the objects, which significantly reduces the regression solution space and improves the convergence speed of the model. Finally, we jointly decode the predicted components, enhancing the internal relationship between the size components and improving the accuracy consistency. Experiments show that when using CenterNet as the improved baseline and Hourglass-104 as the backbone, on the MS COCO dataset, our improved model achieved 44.7%, which is 2.6% higher than the baseline.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 2006.05a
• /
• pp.185-186
• /
• 2006

Image Guided Surgery(IGS) system has been developed to provide exquisite and objective information to surgeons for surgical operation process. It is necessary that registration technique is important to match between 3D image model reconstructed from image modalities and the object operated by surgeon. Majority techniques of registration in IGS system have been used by recognizing fiducial markers placed on the object. However, this method has been criticized due to its invasive protocol inserting fiducial markers in patient's bone. Therefore, shape-based registration technique using geometric characteristics of the object has been invested to improve the limitation of IGS system. During Total Knee Replacement(TKR) operation, it is challenge to register with high accuracy by using shape-based registration because the area to acquire sample data from knee is limited. We have developed region-based 3D registration technique based on anatomical landmarks on the object and this registration algorithm was evaluated in femur model. It was found that region-based algorithm can improve the accuracy in 3D registration. We expect that this technique can efficiently improve the IGS system.

• Smart Structures and Systems
• /
• v.24 no.6
• /
• pp.745-757
• /
• 2019

Road surface deteriorations such as potholes have caused motorists heavy monetary damages every year. However, effective road condition monitoring has been a continuing challenge to road owners. Depth cameras have a small field of view and can be easily affected by vehicle bouncing. Traditional image processing methods based on algorithms such as segmentation cannot adapt to varying environmental and camera scenarios. In recent years, novel object detection methods based on deep learning algorithms have produced good results in detecting typical objects, such as faces, vehicles, structures and more, even in scenarios with changing object distances, camera angles, lighting conditions, etc. Therefore, in this study, a Deep Learning Pothole Detector (DLPD) based on the deep region-based convolutional neural network is proposed for autonomous detection of potholes from images. About 900 images with potholes and road surface conditions are collected and divided into training and testing data. Parameters of the network in the DLPD are calibrated based on sensitivity tests. Then, the calibrated DLPD is trained by the training data and applied to the 215 testing images to evaluate its performance. It is demonstrated that potholes can be automatically detected with high average precision over 93%. Potholes can be differentiated from manholes by training and applying a manhole-pothole classifier which is constructed using the convolutional neural network layers in DLPD. Repeated detection of the same potholes can be prevented through feature matching of the newly detected pothole with previously detected potholes within a small region.

• The Journal of Korean Association of Computer Education
• /
• v.17 no.2
• /
• pp.107-113
• /
• 2014

In this study, we design and implement an eye-gaze estimation system based on the extraction of eye contour and pupil region. In order to effectively extract the contour of the eye and region of pupil, the face candidate regions were extracted first. For the detection of face, YCbCr value range for normal Asian face color was defined by the pre-study of the Asian face images. The biggest skin color region was defined as a face candidate region and the eye regions were extracted by applying the contour and color feature analysis method to the upper 50% region of the face candidate region. The detected eye region was divided into three segments and the pupil pixels in each pupil segment were counted. The eye-gaze was determined into one of three directions, that is, left, center, and right, by the number of pupil pixels in three segments. In the experiments using 5,616 images of 20 test subjects, the eye-gaze was estimated with about 91 percent accuracy.

• Journal of Broadcast Engineering
• /
• v.22 no.5
• /
• pp.608-617
• /
• 2017

Methods of contrast enhancement have problem such as side effect of over-enhancement with non-gaussian histogram distribution, tradeoff enhancement efficiency against brightness preserving. In order to enhance contrast at various histogram distribution, segmentation to region with gaussian distribution and then enhance contrast each region. First, we segment an image into several regions using GMM(Gaussian Mixture Model)fitting by that k-mean clustering and EM(Expectation-Maximization) in $L^*a^*b^*$ color space. As a result region segmentation, we get the region map and probability map. Then we apply local contrast enhancement algorithm that mean shift to minimum overlapping of each region and preserve brightness histogram equalization. Experiment result show that proposed region based contrast enhancement method compare to the conventional method as AMBE(AbsoluteMean Brightness Error) and AE(Average Entropy), brightness is maintained and represented detail information.

• Journal of the Korea Convergence Society
• /
• v.6 no.3
• /
• pp.51-57
• /
• 2015

This paper proposes an efficient method such as Extended Three Region Partitioning Method for nested loops with irregular dependences for maximizing parallelism. Our approach is based on the Convex Hull theory, and also based on minimum dependence distance tiling, the unique set oriented partitioning, and three region partitioning methods. In the proposed method, we eliminate anti dependences from the nested loop by variable renaming. After variable renaming, we present algorithm to select one or more appropriate lines among given four lines such as LMLH, RMLH, LMLT and RMLT. If only one line is selected, the method divides the iteration space into two parallel regions by the selected line. Otherwise, we present another algorithm to find a serial region. The selected lines divide the iteration space into two parallel regions as large as possible and one or less serial region as small as possible. Our proposed method gives much better speedup and extracts more parallelism than other existing three region partitioning methods.

• Journal of Broadcast Engineering
• /
• v.18 no.3
• /
• pp.393-400
• /
• 2013

Hand shape recognition is a fundamental technique for implementing natural human-computer interaction. In this paper, we discuss a method for effectively detecting a hand region in Kinect-based hand shape recognition. Since Kinect is a camera that can capture color images and infrared images (or depth images) together, both images can be exploited for the process of detecting a hand region. That is, a hand region can be detected by finding pixels having skin colors or by finding pixels having a specific depth. Therefore, after analyzing the performance of each, we need a method of properly combining both to clearly extract the silhouette of hand region. This is because the hand shape recognition rate depends on the fineness of detected silhouette. Finally, through comparison of hand shape recognition rates resulted from different hand region detection methods in general environments, we propose a high-performance hand region detection method.

• Archives of Plastic Surgery
• /
• v.41 no.6
• /
• pp.673-678
• /
• 2014

Background The depth of graft harvest and the residual dermis available for reepithelization primarily influence the healing of split-skin graft donor sites. When the thigh region is chosen, the authors hypothesize based on thickness measurements that the anterolateral region is the optimal donor site. Methods Full-thickness skin specimens were sampled from the anteromedial, anterior, and anterolateral regions of human cadavers. Skin specimens were cut perpendicularly with a custom-made precision apparatus to avoid the overestimation of thickness measurements. The combined epidermal and dermal thicknesses (overall skin thickness) were measured using a digital calliper. The specimens were histologically stained to visualize their basement membrane, and microscopy images were captured. Since the epidermal thickness varies across the specimen, a stereological method was used to eliminate observer bias. Results Epidermal thickness represented 2.5% to 9.9% of the overall skin thickness. There was a significant difference in epidermal thickness from one region to another (P<0.05). The anterolateral thigh region had the most consistent and highest mean epidermal thickness ($60{\pm}3.2{\mu}m$). We observed that overall skin thickness increased laterally from the anteromedial region to the anterior and anterolateral regions of the thigh. The overall skin thickness measured $1,032{\pm}435{\mu}m$ in the anteromedial region compared to $1,220{\pm}257{\mu}m$ in the anterolateral region. Conclusions Based on skin thickness measurements, the anterolateral thigh had the thickest epidermal and dermal layers. We suggest that the anterolateral thigh region is the optimal donor site for split-skin graft harvests from the thigh.

• Journal of Broadcast Engineering
• /
• v.6 no.1
• /
• pp.108-118
• /
• 2001

In this paper, the conventional deblocing algorithms are compared in each region of the image. Based on the comparison result, we propose a new deblocklng algorithm that can improve subjectives as well as the objective quality. Because the human visual system is mode sensitive to the blocking artifacts in the low frequency re91on, we compare the performance of several deblocking algorithms in 7he low and high frequency region separately. For this purpose we also propose an algorithm for classifying the region into low and high frequency ones. and propose a deblocking algorithm which is applied differently to each region. The result shows that the adaptive LPF method yields the best performance in the low frequency region in terms of both subjective and objective quality. Hence. by applying the adaptive LPF method to the low frequency region, the performance of conventional algorithms can be improved. In the high frequency region. it is observed that the DCT-based POCS algorithm provides the best performance. Hence, by combining the algorithm with the adaptive LPF method, the best objective performance is obtained.