• Annual Conference of KIPS
• /
• 2024.10a
• /
• pp.867-868
• /
• 2024

도로의 노후화로 포트홀 발생 빈도가 늘어나고 그로 인해 발생하는 사고와 피해액 또한 수억 원대에 이르고 있다. 포트홀 문제를 해결하기 위해 AI 기술, 특히 Image Processing을 이용한 기술이 많이 사용되고 있다. 하지만 실시간으로 영상을 촬영하면서 포트홀을 감지하는 방식은 많은 데이터 처리량과 비용 문제로 인해 일반 운전자들이 쉽게 사용할 수 없다. 본 연구에서는 이러한 문제를 해결하기 위해 두 단계로 나누어 포트홀을 감지하는 방법을 제시한다. 첫 번째 단계에서는 도로 주행 시의 3축 기울기 변화량 값을 분석하여 포트홀임을 구분하고 해당 위치의 사진을 GPS 좌표와 함께 촬영 및 저장한다. 두 번째 단계에서는 촬영한 사진으로부터 OpenCV의 Yolov5를 이용하여 딥러닝을 통한 포트홀을 감지한다. 제안한 시스템으로 데이터 처리량을 줄이고 비용을 절감해 많은 운전자들에게 포트홀 안전 시스템을 보급화 할 수 있다.

• Journal of Broadcast Engineering
• /
• v.22 no.6
• /
• pp.724-733
• /
• 2017

Conventional methods of automatically recording Go game positions do not properly consider obstacles (hand or object) on a Go board during the Go game. If the Go board is blocked by obstacles, the position of a Go stone may not be correctly recognized, or the sequences of moves may be stored differently from the actual one. In the proposed algorithm, only the complete Go board image without obstacles is stored as a background image and the obstacle is recognized by comparing the background image with the current input image. To eliminate the phenomenon that the shadow is mistaken as obstacles, this paper proposes the new obstacle detection method based on the gradient image instead of the simple differential image. When there is no obstacle on the Go board, the background image is updated. Finally, the successive background images are compared to recognize the position and type of the Go stone. Experimental results show that the proposed algorithm has more than 95% recognition rate in general illumination environment.

• The Journal of the Korea Contents Association
• /
• v.7 no.2
• /
• pp.21-29
• /
• 2007

Anisotropic diffusion is one of the widely used techniques in the field of image segmentation. In the conventional anisotropic diffusion [1]-[6], usually 4-neighborhood directions are used: north, south, west and east, except the image diagonal directions, which results in the loss of image details and causes false contours. Existing methods for image segmentation using conventional anisotroplc diffusion can't preserve contour information, or noises with high gradients become more salient as the umber of times of the diffusion increases, resulting in over-segmentation when applied to watershed. In this paper, to overcome the shortcoming of the conventional anisotropic diffusion method, a new arusotropic diffusion method based on diagonal edges is proposed. And a method of watershed segmentation is applied to the proposed method. Experimental results show that the process time of the proposed method including diagonal edges over conventional methods can be up to 2 times faster and the Circle image quality improvement can be better up to $0.45{\sim}2.33(dB)$. Experiments also show that images are segmented very effectively without over segmentation.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.38C no.5
• /
• pp.409-419
• /
• 2013

Kinect released by Microsoft in November 2010 is a motion sensing camera in xbox360 and gives depth and color images. However, Kinect camera also generates holes and noise around object boundaries in the obtained images because it uses infrared pattern. Also, boundary flickering phenomenon occurs. Therefore, we propose a real-time virtual-view video synthesis algorithm which results in a high-quality virtual view by solving these problems. In the proposed algorithm, holes around the boundary are filled by using the joint bilateral filter. Color image is converted into intensity image and then flickering pixels are searched by analyzing the variation of intensity and depth images. Finally, boundary flickering phenomenon can be reduced by converting values of flickering pixels into the maximum pixel value of a previous depth image and virtual views are generated by applying 3D warping technique. Holes existing on regions that are not part of occlusion region are also filled with a center pixel value of the highest reliability block after the final block reliability is calculated by using a block based gradient searching algorithm with block reliability. The experimental results show that the proposed algorithm generated the virtual view image in real-time.

• Journal of the Korean Society of Radiology
• /
• v.5 no.4
• /
• pp.161-169
• /
• 2011

This research experimented on the change of the multiple colleague scan angle facing one scan object facet to many directions of the form of 3D about the visual angle nervous system forming the cubic distribution with the gradient magnetic field of the mri system and considered the existing basic angle oblique direction test coverage and comparison. MR system can freely select various pulse sequence and image slice. To oblique imaging for optic nerve viewing, we have studied the variation of scan angle between typical oblique scan method (sagittal-coronal plane) and multi directional angles oblique scan method (sagittal-coronal-axial plane) using gradient of MR system. In this study, the subjects of the experiment were normal adults in our country. As a result, we confirmed that multi directional angles oblique scan method can display anatomical information of more wider area than typical oblique scan method. In addition, to clearly display optic nerve, we also confirmed that image slice thickness and pulse sequence have effect on it.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.50 no.4
• /
• pp.163-170
• /
• 2013

In general, images of outdoor scenes often contain degradation due to dust, water drop, haze, fog, smoke and so on, as a result they cause the contrast reduction and color fading. Haze removal is not easier problem due to the inherent ambiguity between the haze and the underlying scene. So, we propose a novel method to solve single scene dehazing problem using the region segmentation based on graph algorithm that has used a gradient value as a cost function. We segment the scene into different regions according to depth-related information and then estimate the global atmospheric light. The medium transmission can be directly estimated by the threshold function of graph-based segmentation algorithm. After estimating the medium transmission, we can restore the haze-free scene. We evaluated the degree of the visibility restoration between the proposed method and the existing methods by calculating the gradient of the edge between the restored scene and the original scene. Results on a variety of outdoor haze scene demonstrated the powerful haze removal and enhanced image quality of the proposed method.

• Journal of Broadcast Engineering
• /
• v.20 no.2
• /
• pp.238-247
• /
• 2015

We develop the signal processing method for adaptive implementing direction of signal and the frequency sensitivity of human visual system(HVS). Existing multibnad energy scaling method makes ringing artifact because it does not consider signal direction. To solve this problem, we use block gradient for signal direction in addition to existing method. And we use the fact that frequency component of signal is more sensitive than value of signal over human eyes. we enhance the signal according to contrast sensitivity function(CSF) which is the model of frequency sensitivity of human eye. Compared that the existing analysis models only improve the efficiencies in the existing systems, the developed method can process the image signals to be more desirable and suitable to HVS.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.20 no.3
• /
• pp.621-629
• /
• 2016

This paper presents a decision method of middle ear disease which is developed in children and adults. In the proposed method, features are extracted from the middle ear disease images and normal images using HoG (histogram of oriented gradient) descriptor and the extracted features are learned by SVM (support vector machine) classifier. To obtain an input vector into SVM, an input image is resized to a predefined size and then the resized image is partitioned into 16 blocks each of which is partitioned into 4 sub-blocks (namely cell). Finally, the feature vector with 576 components is given by using HoG with 9 bins and it is used as SVM learning and classification. Input images are classified by SVM classifier based on the model of learning features. Experimental results show that the proposed method yields the precision of over 90% in decision.

• The Transactions of the Korea Information Processing Society
• /
• v.3 no.5
• /
• pp.1337-1345
• /
• 1996

Edge detection is the first step and very important step in image nalyisi. In this paper, proposed edge detection algorithm based on informations of edge structures and it is different from other classical edge detection operators such asgradient and surface fitting algorithm. The firs, we defined characteristics of edge structures such as continuity, thinness, localization, length. The second, we defined valid edge structures and ideal edge pixel positions in $3\times3$ window based on edge characteristics of edge structures. And we proposed twelve windows for enhance dissimilarity regions based on valid edge structures and ideal edge pixel positions. In specially, proposed algorithm was shown better performance of edge detection than other operators such as gradient operator and the LoG(Laplacian of gradient) operator of zero crossings in noisy test image with $\sigma=30$.

• Journal of Broadcast Engineering
• /
• v.23 no.2
• /
• pp.206-217
• /
• 2018

We develop the signal processing method for adaptive implementing direction of signal and the frequency sensitivity of human visual system(HVS). Existing multiband energy scaling method makes ringing artifact because it does not consider signal direction. To solve this problem, we use block gradient for signal direction in addition to existing method. And we use the fact that frequency component of signal is more sensitive than value of signal over human eyes. we enhance the signal according to contrast sensitivity function(CSF) which is the model of frequency sensitivity of human eye. Compared that the existing analysis models only improve the efficiencies in the existing systems, the developed method can process the image signals to be more desirable and suitable to HVS.