• 한국전산구조공학회논문집
• /
• 제35권5호
• /
• pp.259-266
• /
• 2022

본 논문에서는 프리팹 구조물의 품질관리를 위한 딥러닝 및 비전센서 기반의 조립 성능 평가 모델을 개발하였다. 조립부 검출을 위해 인코더-디코더 형식의 네트워크와 수용 영역 블록 합성곱 모듈을 적용한 딥러닝 모델을 사용하였다. 검출된 조립부 영역 내의 볼트홀을 검출하고, 볼트홀의 위치 값을 산정하여 k-근접 이웃 기반 모델을 사용하여 조립 품질을 평가하였다. 제안된 기법의 성능을 검증하기 위해 조립부 모형을 3D 프린팅을 이용하여 제작하여 조립부 검출 및 조립 성능 예측 모델의 성능을 검증하였다. 성능 검증 결과 높은 정밀도로 조립부를 검출하였으며, 검출된 조립부내의 볼트홀의 위치를 바탕으로 프리팹 구조물의 조립 성능을 5% 이하의 판별 오차로 평가할 수 있음을 확인하였다.

• Journal of Information Processing Systems
• /
• 제8권2호
• /
• pp.359-374
• /
• 2012

In this paper, we propose a new machine vision algorithm for automatic defect detection on patterned textures with the help of texture-periodicity and the Jensen-Shannon Divergence, which is a symmetrized and smoothed version of the Kullback-Leibler Divergence. Input defective images are split into several blocks of the same size as the size of the periodic unit of the image. Based on histograms of the periodic blocks, Jensen-Shannon Divergence measures are calculated for each periodic block with respect to itself and all other periodic blocks and a dissimilarity matrix is obtained. This dissimilarity matrix is utilized to get a matrix of true-metrics, which is later subjected to Ward's hierarchical clustering to automatically identify defective and defect-free blocks. Results from experiments on real fabric images belonging to 3 major wallpaper groups, namely, pmm, p2, and p4m with defects, show that the proposed method is robust in finding fabric defects with a very high success rates without any human intervention.

• Journal of the Optical Society of Korea
• /
• 제19권6호
• /
• pp.687-694
• /
• 2015

A photoluminescence (PL) imaging method using a vision camera was employed to inspect InGaN/GaN quantum-well light-emitting diode (LED) epi-wafers. The PL image revealed dark spot defective regions (DSDRs) as well as a spatial map of integrated PL intensity of the epi-wafer. The Shockley-Read-Hall (SRH) nonradiative recombination coefficient increased with the size of the DSDRs. The high nonradiative recombination rates of the DSDRs resulted in degradation of the optical properties of the LED chips fabricated at the defective regions. Abnormal current-voltage characteristics with large forward leakages were also observed for LED chips with DSDRs, which could be due to parallel resistances bypassing the junction and/or tunneling through defects in the active region. It was found that the SRH nonradiative recombination process was dominant in the voltage range where the forward leakage by tunneling was observed. The results indicated that the DSDRs observed by PL imaging of LED epi-wafers were high density SRH nonradiative recombination centers which could affect the optical and electrical properties of the LED chips, and PL imaging can be an inspection method for evaluation of the epi-wafers and estimation of properties of the LED chips before fabrication.

• Journal of Biosystems Engineering
• /
• 제26권3호
• /
• pp.271-278
• /
• 2001

Many workers in the garlic peeling factory are separating the sound peeled garlics from the unpeeled and defective ones in a manual way. In order to reduce the seasonal labor requirement and operating cost, the mechanized garlic sorting system such as the vision-based garlic sorter should be developed. This study was conducted as one of basic studies on developing quality evaluation technique for peeled garlics, especially to developed the system for acquiring the whole surface images of garlics with a CCD camera. The following results were obtained from this study. 1. The belt-type garlic rotation system was devised to apply for the vision-based garlic sorter and was tested to decide the criteria of design and optimum conveying speed. 2. To evaluate the performance of the developed garlic rotation system, feeding rate and rotating rate were measured under the conditions of four experimental factors such as the inclined angle of rotating belt, the inclined angle of feeding belt, the height of plate arrays on feeding belt and the conveying speed of belts. And the capacity of the system according to mixture ratios of peeled garlics and unpeeled garlics was analyzed as a feasibility test. 3. For the inclined angle of rotating belt 20°and height of plate array on feeding belt 22㎜, the maximum rotating rate for garlic samples including unpeeled ones was 81.1% at the conveying speed of 4.2 garlic/sec. And under these condition, the maximum feeding rate was 85% at the inclined angle of feeding belt 6.5°. 4. The capacity of the developed garlic rotation system was almost constant regardless of mixture ratio of peeled garlics and unpeeled garlics and its range was 2.95∼3.92 garlic/sec. At the conveying speed of 4.2 garlic/sec, the capacity of the garlic rotation system was calculated ad 58∼64 kg/hr. 5. To improve performance of the garlic rotation system, it is recommended to develop a device to slide garlics into feeding belt.

• 한국멀티미디어학회논문지
• /
• 제23권11호
• /
• pp.1383-1387
• /
• 2020

Recently, a 4-col 7-row LED Screen that provides various information of major roads and local governments has been installed and operated. However, due to deterioration due to changes in temperature and humidity, deterioration due to static electricity, and mechanical stress, partial module failure of the display may occur, which is a major cause of missing information of vitally given to citizens. However, there have been frequent cases where the 4-col and 7-row LED Screen that have failed due to reasons such as installed location where the signboards are installed on the road and outdoor, the lack of monitoring means at all times, and the lack of manpower is often neglected for a long time. Following this flow, this paper proposes a method to detect defective modules by analyzing the images collected through the camera fixed to the front part of the LED display.

• 반도체디스플레이기술학회지
• /
• 제12권1호
• /
• pp.15-21
• /
• 2013

In this paper, sensor system and inspection algorithm in order to inspect steel balance weight for automobile is described. Steel balance weight is composed of clip and weight, which is joined by press process. The defective one has a gap between clip and weight. To detect whether there is a gap, sensor system is simply configured with laser line projector and USB camera, which make it possible to measure the height difference of clip and weight area. Laser line pattern which is made on the surface of a balance weight is captured by USB camera. In case that USB camera is used in machine vision, barrel distortion caused by wide angle lens makes the captured image distorted. Image warping function is applied to correct the distortion. Simple image processing algorithm is applied to extract the laser line information and whether it is good or not is judged through the extracted information.

• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2022년도 춘계학술대회
• /
• pp.108-111
• /
• 2022

제품의 결함 탐지를 위한 머신 비전 시스템에 딥러닝을 적용하기 위해서는 다양한 결함 사례에 대한 방대한 학습 데이터가 필요하다. 하지만 실제 제조 산업에서는 결함의 종류에 따른 데이터 불균형이 생기기 때문에 결함 사례를 일반화할 수 있을 만큼의 제품 이미지를 수집하기 위해서는 많은 시간이 소요된다. 본 논문에서는 적은 데이터로도 학습이 가능한 샴 신경망을 제품 결함 탐지에 적용하고, 제품 결함 이미지 데이터의 속성을 고려하여 이미지 쌍 구성법과 대조 손실 함수를 수정하였다. AUC-ROC로 샴 신경망의 임베딩 성능을 간접적으로 확인한 결과, 같은 제품끼리만 쌍을 구성하고 결함이 있는 제품 간에는 쌍을 구성하였을 때, 그리고 지수 대조 손실로 학습하였을 때 좋은 임베딩 성능을 보였다.

• 한국농업기계학회:학술대회논문집
• /
• 한국농업기계학회 2000년도 THE THIRD INTERNATIONAL CONFERENCE ON AGRICULTURAL MACHINERY ENGINEERING. V.III
• /
• pp.551-559
• /
• 2000

An integrated on-line inspection system was constructed with seven cameras, half mirrors to split images, 720 nm and 970 nm band pass filters, illumination chamber having several tungsten-halogen lamps, one main computer, one color frame grabber, two 4-channel multiplexors, and flat plate conveyer, etc., so that a total of seven images, that is, one color image from the top side of an apple and two B/W images from each side (top, right and left) could be captured and displayed on a computer monitor through the multiplexor. One of the two B/W images captured from each side is 720nm filter image and the other is 970nm. With this system an on-line grading software was developed to evaluate appearance quality. On-line test results to the Fuji apples that were manually fed on the conveyer showed that grading accuracies of the color, defective and shape were 95.3%, 86% and 91%, respectively. Grading time was 0.35 sec per apple on an average. Therefore, this on-line grading system could be used for inspection of the final products produced from an apple sorting system.

• 감성과학
• /
• 제16권4호
• /
• pp.557-566
• /
• 2013

본 연구는 색각이상자들의 색채 감성 연상의 실태 조사 목적에서 기획되었다. 정상인 100명과 색각이상자 34명으로 피실험 집단을 구성하였고, 색각이상자를 '적색약' 집단 8명과 '녹색약' 16명으로 세분하여, 좋아하는 색, 행복한 색, 친근한 색, 싫어하는 색, 슬픈 색, 거북한 색, 적극적인 색과 소극적인 색 등 항목에 대해서 먼셀이 제안한 기본 10색 가운데 하나씩만 선택하도록 하였다. 정상인 집단의 좋아하는 색은 파랑과 빨강이고, 행복한 색은 노랑이며, 친근한 색은 초록이었다. 색각이상자 집단은 좋아하는 색으로 파랑을, 행복한 색으로 노랑을, 친근한 색으로 파랑을 선택하였다. 또한 정상인 집단이 싫어하는 색을 청록으로, 슬픈 색으로 파랑을, 거북한 색으로 청록을 선택한 것과 비교하여, 색각이상자 집단은 싫어하는 색으로 청록을, 슬픈 색으로 자주를, 거북한 색으로 청록을 선택하였다. 정상인 집단이 적극적인 색과 소극적인 색으로 빨강과 청록을 선택한 반면, 색각이상자 집단은 각각 빨강과 파랑을 선택하였다. 이는 색각이상자들의 일상생활에 있어서 색채의 왜곡된 수용이 감성왜곡으로 연결되어 일상생활에서의 부정적 요소가 되지 않도록 유도하는 색 구성 작업에 활용되고, 특히 문화콘텐츠 이용 편의성 등에 활용될 수 있을 것이다.

• Design & Manufacturing
• /
• 제14권2호
• /
• pp.14-20
• /
• 2020

In OLED panel production process, it is necessary to cut a part of protective film as a preprocess for lighting inspection. The current method is to recognize only the fiducial mark of the cut-out panel. Bare Glass Cutting does not compensate for machining cumulative tolerances. Even though process defects still occur, it is necessary to develop technology to solve this problem because only the Align Mark of the panel that has already been cut is used as the reference point for alignment. There is a lot of defective lighting during panel lighting test because the correct protective film is not cut on the panel power and signal application pad position. In laser cutting process to remove the polarizing film / protective film / TSP film of OLED panel, laser processing is not performed immediately after the panel alignment based on the alignment mark only. Therefore, in this paper, we performed real time inspection which minimizes the mechanism tolerance by correcting the laser cutting path of the protective film in real time using Machine Vision. We have studied calibration algorithm of Vision Software coordinate system and real image coordinate system to minimize inspection resolution and position detection error and edge detection algorithm to accurately measure edge of panel.