• Proceedings of the KIEE Conference
• /
• 2006.07d
• /
• pp.2049-2050
• /
• 2006

최근 제품의 품질 향상과 생산력 제고를 위해 제조업 분야에서의 결함 검출 시스템의 자동화에 대한 필요성이 커지고 있다. 또한 많은 회사들이 자동 흠 검출 시스템을 도입하여 판재나 모직 등의 흠 검출에 사용하고 있다. 그러나 철강 제조업에서 선재의 흠 검출은 이전에 많이 시도된바 있는 판재 등의 흠 검출보다 극복해야할 과제가 많다. 우선 선재의 특성상 조명이 일정하더라도 각 부분의 밝기가 다르고, 선재가 이동하거나 선재의 굵기가 달라지면 카메라에서 관측되는 부위의 밝기가 달라진다. 따라서 선재의 흠 검출 시스템에서는 이들의 영향을 극복하고 제품의 결함을 판별할 수 있는 알고리즘이 요구된다. 또한 선재의 이동속도는 최고 18m/s에 이르기 때문에 실시간으로 흠을 검출하기 위해서는 알고리즘이 간단하고 효율적이어야 한다. 본 논문에서는 이러한 문제를 극복하고 선재의 흠을 자동 검출하기 위한 알고리즘을 제시한다.

• The KIPS Transactions:PartB
• /
• v.13B no.4 s.107
• /
• pp.377-382
• /
• 2006

In this paper, two methods based on the use of morphology and pattern matching prior to detect classified defects automatically on the back light unit which is a part of display equipments are proposed. One is the template check method which detects small size defects by using closing and opening method, and the other is the block matching method which detects big size defects by comparing with four regions of uniform blocks. The TC algorithm also can detect defects on the non-uniform pattern of BLU by using revised Otsu method. The proposed method has been implemented on the automatic defect detection system we developed and has been tested image data of BLU captured by the system.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.17 no.11
• /
• pp.2486-2491
• /
• 2013

There are invisible defects by naked eyes in ceramic material images such as internal stomata, cracks and foreign substances. In this paper we propose a method to detect and extract such defects from ceramic pipe weld zone by applying ART2 learning. In pre-processing, we apply Ends-in Search Stretching to enhance the intensity and then perform fuzzy binarization with triangle type membership function followed by enhanced ART2 that interacts with random input patterns to extract such invisible defects. The experiment verifies that this proposed method is sufficiently effective.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2016.06a
• /
• pp.300-301
• /
• 2016

본 논문에서는 동영상에서 제한된 종류의 동적 객체를 자동적으로 검출하여 추적하는 기법을 제안한다. 제안하는 기법은 객체 검출 기법[1]과 객체 추적 기법[2]의 협업을 통해 이를 수행한다. 검출기는 매 장면마다 객체들을 검출하고 이 중 높은 신뢰도의 객체에 대해 추적을 시작한다. 추적기는 이전 장면에서 학습된 분류기에 기반하여 객체를 추적한다. 추적 결과와 겹치는 검출 결과를 분석하여 현재 장면에서 객체의 정확한 위치와 모양을 추정한다. 겹치는 검출 결과가 없을 때는 검출기로 부터 추적 결과의 신뢰도를 측정하고 문턱값에 따라 추적을 계속 진행하거나 종료한다. 실험 결과를 통해 제안하는 기법이 기존 검출 기법에 비해 우수한 검출 성능을 보임을 확인한다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2016.11a
• /
• pp.21-22
• /
• 2016

본 논문에서는 동영상에서 객체를 자동 검출하는 기법을 제안한다. 제안하는 기법은 정지 영상에서 객체를 검출하는 기법과 동영상에서 객체를 추적하는 기법을 동시에 수행하여 동영상에서 객체를 검출한다. 매 프레임 검출기는 학습된 종류의 객체들을 검출하고 추적기는 이전 프레임에서 검출되었던 객체를 추적한다. 검출기가 검출한 결과와 추적기가 추적한 결과를 매칭하고, 겹치는 결과와 그렇지 않은 결과에 대해 각각 다른 검사를 수행하여 신뢰도 있는 결과를 도출한다. 실험 결과를 통해 제안하는 기법이 기존 검출 기법에 비해 우수한 성능을 보임을 확인한다.

• Korean Journal of Optics and Photonics
• /
• v.10 no.2
• /
• pp.114-119
• /
• 1999

In an interation-free measurement system with the polarization-based Mach-Zehnder interferometer structure, effects of two system imperfections, i.e., the system loss and the low quantum efficiency of photon detectors, on the system perfomance are analyzed, Although both the two system imperfections limit the maximal defection efficiency that can be achievable, we show that the quantum efficiency of photon detectors is less important when the system loss becomes small. In addition, for given system losses and quantum efficiencies, we derive the maximal detection efficiencies that can be achievable.

• Journal of the Korean Society for Nondestructive Testing
• /
• v.17 no.1
• /
• pp.23-30
• /
• 1997

Recently high frequency ultrasonic transducers to employ polyvinylidene fluoride(PVDF) or polyvinylidene fluoride trifluoroethylene P(VDF-TrFE) have been used to detect small flaws in immersion testing. The detection field depending on the water path between the transducer and a specimen and the path in a tested specimen was measured using a PVDF transducer with nominal frequency 80MHz. Also, C-scan and B-scan were performed for the specimens made of Cr-Ni steel with the artificial flaws, the flat-bottom holes with diameter ranging from $50{\mu}m$ to $560{\mu}m$ at 12mm depth. As the result, the flaws with diameter larger than $280{\mu}m$ were detected, but the flaws with the ratio of diameter to wavelength smaller than about 0.48 were not detected. That the smaller flaws could not be detected was attributed to the attenuation of high frequency components in the steel specimens.

• Journal of Sensor Science and Technology
• /
• v.10 no.1
• /
• pp.33-41
• /
• 2001

Wire ropes have been widely used in industrial applications, wherever heavy weight should be carried safely or mechanical energy should be transmitted fast. Especially, wire rope failures in operating elevator may lead to extensive property damage and serious injury to nearby personnel. Hence, it is very important to inspect wire rope periodically. Failure defection of wire rope requires fundamental knowledge of wire rope construction, rope behavior, properties of fault, sensing and signal processing method. In this research, the development of a new fault detecting system incorporating Hall-effect sensors to detect flaws such as abrasion, broken wire, corrosion and deformation for aged wire ropes in elevator, is described. For using a detector as a portable instrument, several performances for implementing sensing part with Hall-effect sensor, analog signal processing unit and programs are described. Experiments and field testing results for the implemented detecting system are also given. As a result, it is verified that the detecting system has good efficiency for inspecting faults of aged wire ropes in service.

• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• v.10 no.1
• /
• pp.47-60
• /
• 2006

Due to the uneven illumination of FPD panel surface, it is difficult to detect the defects. The paper proposes a method to find the uneven illumination compensation using wavelets, which are done based on multi-resolution structure. The first step is to decompose the image into multi-resolution levels. Second, elimination of lowest smooth sub-image with highest frequency band removes the high frequency noise and low varying illumination. In particular, the main algorithm was implemented by lifting scheme for realtime inline process.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.10 no.3
• /
• pp.529-534
• /
• 2009

In this paper, we propose a new algorithm which can detect Mura in TPT-LCD effectively. Since Mura in TFT-LCD has a certain area shape, it is seen as a sin wave in a LCD line. Consequently, it is shown that this type of Mura can be detected easily through FFT. Even multiple size of Mura patterns exist, those patterns can be detected with multi-point FFT. Proposed algorithm can be utilized in automatic Mura detection systems instead of human Mura detection methods.