• Proceedings of the Korean Institute Of Construction Engineering and Management
• /
• autumn
• /
• pp.535-538
• /
• 2002

Machine vision algorithms, which are composed of noise elimination algorithm, crack detection and mapping algorithm, and path planning algorithm, are required for sealing crack networks effectively and automation of crack sealing.. Noise elimination algorithm is the first step so that computer take cognizance of cracks effectively. Noises should be removed because common road includes a lot of noises(mark of oil, tire, traffic lane, and sealed crack) that make it difficult the computer to acknowledge cracks accurately. The objective of this paper is to propose noise elimination algorithm, prove the efficiency of the algorithm through coding. The result of the coding is represented in this paper as well.

• KSCI Review
• /
• v.15 no.1
• /
• pp.173-177
• /
• 2007

3D 디스플레이 기술이 발전함에 따라 스테레오 영상의 전송시 요구되는 비트레이트의 감소가 절실히 필요하다. 하지만, 스테레오 영상은 서로 다른 카메라에 의해 취득되기 때문에 잠재적으로 서로 차이가 있고, 이것은 디스패리티 추정시 큰 오차를 유발할 수 있으며 전송될 비트레이트에 영향을 줄 수 있다. 따라서 스테레오 영상들 사이의 밸런싱이 필요하다. 스테레오 영상의 밸런싱을 위해, 본 논문에서는 히스토그램 Specification 방법과 타깃 영상의 국부정보, 스테레오 영상간의 오차 분포를 이용한다. 히스토그램 Specification 방법은 그레이레벨의 맵핑관계를 정의한다. 따라서 이를 통해 맵핑될 레벨의 맵핑 구간을 구할 수 있다. 그 구간에서, 맵핑될 기준영상의 히스토그램 분포와 스테레오 오차값의 분포는 서로 모양이 유사할 것이다. 그러나, 폐색된 영역이나 노이즈에 의해 그 모양이 변하므로 우리는 맵핑될 픽셀들을 오차영상에서 그 픽셀들의 근방에서 구한 평균들과 오른쪽 영상(타깃 영상)에서 맵핑될 픽셀의 근방에서 구한 평균이 최소 값을 갖는 위치 값으로 맵핑한다. 제안된 방법은 실험에서 기존 방법보다 향상된 결과를 나타내는 것을 보여 준다.

• Transactions of the Korean Society for Noise and Vibration Engineering
• /
• v.25 no.4
• /
• pp.275-283
• /
• 2015

The aim of this study is to estimate the noise exposure of roadside apartment complex according to the characteristics of apartment complex. The facade noise level of residential buildings and the ground noise level inside apartment complexes were predicted and analyzed using noise mapping based on a computerized noise model. In addition, the correlation analysis between these noise levels and the characteristics of apartment complex such as traffic volume, building coverage, the number of adjacent roads, etc. was done in a total of 21 apartment complexes. The results showed that building facade noise level and ground noise level were positively correlated with traffic volume (correlation coefficient, r=0.616~0.623) and the number of adjacent roads (r=0.340~0.496). On the other hand, they were negatively correlated with building coverage (r=-0.413~-0.477) and complex area per the number of roads (r=-0.478~-0.615).

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2013.07a
• /
• pp.21-22
• /
• 2013

논문에서는 제스처 인식을 하기 위해 필요한 특징 값을 3차원 방향 코드로 구현한 특징 패턴을 검출하는 방법을 제안한다. 검출된 데이터 좌표끼리 직선을 만들고 직선들의 사이각의 합 연산을 이용해서 특징 변곡점을 추출한다. 추출된 변곡점끼리 직선을 생성한 후, 8방향 코드와 깊이 값을 병합시킨 24방향 코드를 맵핑 시켜준다. 맵핑된 방향 코드들을 한 패턴으로 생성한다. 생성된 패턴에서 인식에 불필요한 방향 노이즈를 제거하기 위해 특정 규칙을 적용한 필터링을 적용하여 필터링된 패턴을 추출하게 된다. '배너코드를 이용한 8방향 패턴'과 비교해서 더 효과적인 패턴이 추출됨을 확인하였다.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.37A no.12
• /
• pp.1054-1064
• /
• 2012

HDR (high dynamic range) tone mapping algorithms are used in image processing that reduces the dynamic range of an image to be displayed on LDR (low dynamic range) devices properly. The retinex is one of the tone mapping algorithms to provide dynamic range compression, color constancy, and color rendition. It has been developed through multi-scale methods and luminance-based methods. Retinex algorithms still have drawbacks such as the emphasized noise and desaturation. In this paper, we propose a multi scale tone mapping algorithm for enhancement of contrast, saturation, and noise of HDR rendered images based on visual brightness functions. In the proposed algorithm, HSV color space has been used for preserving the hue and saturation of images. And the algorithm includes the estimation of minimum and maximum luminance level and a visual gamma function for the variation of viewing conditions. And subjective and objective evaluations show that proposed algorithm is better than existing algorithms. The proposed algorithm is expected to image quality enhancement in some fields that require a improvement of the dynamic range due to the changes in the viewing condition.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.22 no.2
• /
• pp.319-325
• /
• 2021

A fine dust measurement using drones is becoming an increasingly common technology, and air pollutants can be identified through dust monitoring in partial industrial areas. A station for measuring fine dust provides information at large construction site offices. On the other hand, it was difficult to check the fine dust in the pollutant source accurately. Therefore, the drone took measurements directly after been placed at the site. While measuring fine dust, monitoring noise occurred due to the influence of the drone's down-wind during landing, but the measurements were similar to the numerical value of the grounded pollution source on the height of 30 m. The field applicability to the study area has limitations in periodic updates using satellite images because the terrain was constantly changing due to considerable flattening fieldwork. Therefore, this study implemented a system that can reflect real-time field information through GIS mapping using drones.

• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 2014.10a
• /
• pp.885-887
• /
• 2014

Outdoor spaces in an apartment complex have been enlarged by the increased underground car parking. It has become accepted as important place for acoustic comfort of resident. This paper attempts to determine the noise exposure to the outdoor area in 21 apartment complexes built within 5 years. The results showed that the average noise level of outdoor area ranged from 37.6dB(A) to 67.2dB(A). And the percentage of areas below the noise level of 55dB(A) range 0.1% to 95.0%. The analysis on correlations shows that the traffic volume and building coverage have significant effects on noise level.

• Korean Journal of Construction Engineering and Management
• /
• v.5 no.2 s.18
• /
• pp.90-105
• /
• 2004

Machines for crack sealing automation have been continually developed since the early 1990's because of the effectiveness of crack sealing that would be able to improve safety, quality and productivity. It has been considered challenging problem to detect crack network in pavement which includes noise (oil marks, skid marks, previously sealed cracks and inherent noise). Moreover, it is required to develop crack network mapping and modeling algorithm in order to accurately inject sealant along to the middle of cut crack network. The primary objective of this study is to propose machine vision algorithms (digital image processing algorithm and path planning algorithm) for fully automated pavement crack sealing. It is anticipated that the effective use of the proposed machine vision algorithms would be able to reduce error rate in image processing for detecting, mapping and modeling crack network as well as improving quality and productivity compared to existing vision algorithms.

• Journal of Intelligence and Information Systems
• /
• v.29 no.2
• /
• pp.115-127
• /
• 2023

Video stabilization is one of the camera technologies that the importance is gradually increasing as the personal media market has recently become huge. For deep learning-based video stabilization, existing methods collect pairs of video datas before and after stabilization, but it takes a lot of time and effort to create synchronized datas. Recently, to solve this problem, unsupervised learning method using only unstable video data has been proposed. In this paper, we propose a network structure that learns the stabilized trajectory only with the unstable video image without the pair of unstable and stable video pair using the Convolutional Auto Encoder structure, one of the unsupervised learning methods. Optical flow data is used as network input and output, and optical flow data was mapped into grid units to simplify the network and minimize noise. In addition, to generate a stabilized trajectory with an unsupervised learning method, we define the loss function that smoothing the input optical flow data. And through comparison of the results, we confirmed that the network is learned as intended by the loss function.