• 한국인터넷방송통신학회논문지
• /
• 제15권5호
• /
• pp.183-192
• /
• 2015

본 논문에서는 통신 판매사에서 사용하는 문서와 이미지 파일에서 개인 정보의 유출을 방지할 수 있는 개인 정보 자동 검색(PIAD, Personal Information Auto Detection) 시스템을 제안한다. 제안하는 시스템은 개인 정보를 포함하는 신분증과 계약서 이미지를 자동으로 검색하고 그 결과를 사용자에게 전달하고, 문서상의 개인 정보 또한 검출할 수 있다. 본 시스템은 빠르고 정확한 검색을 위하여 선별 과정과 분석 과정으로 나뉘고, 분석 과정은 SURF, 침식과 팽창, FindContours 알고리즘들을 사용한다. 제안하는 PIAD 시스템은 272장의 입력 이미지들 중 267장을 선별 및 검출함으로써 98% 이상의 정확도를 보였다.

• 한국컴퓨터정보학회논문지
• /
• 제8권4호
• /
• pp.63-68
• /
• 2003

지문 인식 시스템에서는 대부분 융선의 끝점과 분기점을 이용한다. 그러나 융선을 체인 코드화하여 이것을 이용한 훼손된 지문의 인식에 관해 연구하였다. 처리의 결과와 순서는 다음과 같이 요약 할 수 있다. (1)몇 개의 훼손된 지문획득, (2)전처리(잡음 제거를 위해 중간 값 필터링, 지역 또는 전역 히스토그램 균등화, 팽창과 침식, 세선화와 유사 이미지 제거), (3)최소 자승법 처리 후 융선 재구성, (4)체인 코드 벡터의 분포 계산, (5)전 처리된 지문 이미지에서 각 벡터의 값들은 거의 같은 결과 값을 보였다. 이 결과로부터 우리는 특징점을 이용한 시스템과 결합하여 사용한다면 더욱 성공적인 지문 인식 시스템이 될 것이라 추측한다.

• 한국정보전자통신기술학회논문지
• /
• 제12권1호
• /
• pp.92-98
• /
• 2019

본 논문에서는 몰포러지 연산을 기본으로 하는 몰포러지 신경망(MNN: Morphological Neural Network) 기반 딥러닝 시스템을 제안하였다. 딥러닝에 사용되는 레이어는 몰포러지 레이어, 풀링 레이어, ReLU 레이어, Fully connected 레이어 등이다. 몰포러지 레이어에서 사용되는 연산은 에로전, 다이레이션, 에지검출 등이다. 본 논문에서 새롭게 제안한 MNN은 기존의 CNN(Convolutional Neural Network)을 이용한 딥러닝 시스템과는 달리 히든 레이어의 수와 각 레이어에 적용되는 커널 수가 제한적이다. 레이어 단위 처리시간이 감소하고, VLSI 칩 설계가 용이하다는 장점이 있으므로 모바일 임베디드 시스템에 딥러닝을 다양하게 적용할 수 있다. MNN에서는 제한된 수의 커널로 에지와 형상검출 등의 연산을 수행하기 때문이다. 데이터베이스 영상을 대상으로 행한 실험을 통해 MNN의 성능 및 딥러닝 시스템으로의 활용 가능성을 확인하였다.

• 대한전기학회논문지:전력기술부문A
• /
• 제48권9호
• /
• pp.1167-1174
• /
• 1999

In this paper, we present a new technique for the local decomposition of convex structuring elements for morphological image processing. Local decomposition of a structuring element consists of local structuring elements, in which each structuring element consists of a subset of origin pixel and its eight neighbors. Generally, local decomposition of a structuring element reduces the amount of computation required for morphological operations with the structuring element. A unique feature of our approach is the use of linear integer programming technique to determine optimal local decomposition that guarantees the minimal amount of computation. We defined a digital convex polygon, which, in turn, is defined as a convex structuring element, and formulated the necessary and sufficient conditions to decompose a digital convex polygon into a set of basis digital convex polygons. We used a set of linear equations to represent the relationships between the edges and the positions of the original convex polygon, and those of the basis convex polygons. Further. a cost function was used represent the total processing time required for computation of dilation/erosion with the structuring elements in a decomposition. Then integer linear programming was used to seek an optimal local decomposition, that satisfies the linear equations and simultaneously minimize the cost function.

• 반도체디스플레이기술학회지
• /
• 제11권1호
• /
• pp.79-83
• /
• 2012

We use adaptive method and determine threshold coefficient so that the algorithm could decide a suitable binarization threshold coefficient of the image to detecting a marker; therefore, we solve the light influence on the shadow area and dark region. In order to improve the speed for reducing computation we created Integral Image. The algorithm detects an outline of the image by using canny edge detection for getting damage or obscured markers as it receives the noise removed picture. The strength of the line of the outline is extracted by Hough transform and it extracts the candidate regions corresponding to the coordinates of the corners. Markers extracted using the equation of a straight edge to find the coordinates. By using the equation of straight the algorithm finds the coordinates the corners. of extracted markers. As a result, even if all corners are obscured, the algorithm can find all of them and this was proved through the experiment.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1993년도 한국자동제어학술회의논문집(국제학술편); Seoul National University, Seoul; 20-22 Oct. 1993
• /
• pp.283-287
• /
• 1993

We constructed morphological filter for single sweep records of event-related potential (ERP), especially P300 waveform. By combining 4 basic operations; erosion, dilation, opening and closing, we can derive any desired filters whose property fits the current objectives. The morphological filter for single sweep records of ERP was constructed by taking account of the features of the signal and noise components. The morphological filter has superior properties of separating the signal ancl the noise even existing within a same frequency band. The constructed morphological filter was tested by using simulation data of ERP and then applied to actual ERP data of a normal subject. The results proved that the constructed morphological filter was an appropriate tool for single sweep records of ERP.

• 한국통신학회논문지
• /
• 제18권9호
• /
• pp.1239-1247
• /
• 1993

집합이론을 기본으로하는 수학적 형태학은 영상처리의 여러분야에서 다양하게 이용되어 왔다. 본 논문에서는 수학적 형태학을 이용하는 새로운 이진 영상의 피라미드를 제안하였다. 제안된 방법에서는 이진영상을 특별한 조건을 갖는 구조적 요소(structure element)로 침식 (erosion)하고 침식된 영상을 부표본화(subsampling)를 이 용하여 축소하고, 축소된 영상을 외삽(interpolation)과 침식 할 때와 동일한 구조적요소로 융기(dilation)하여 확장하며, 원 영상과 차를 이용하여 나머지 영상을 구하였다. 이때 사용된 구조적 요소는 부표본화 집합을 융기하였을 경우, 원래의 표본화 집합을 재생할 수 있는 구조적 요소로서, 차이 영상이 0또는 1로 표현되기 위해 충분해야 한다. 또한, 피라미드 구성에 필요한 연산들이 분리가능(separable)함을 증명함으로써 파이프라인 프로세싱등 병렬 연간이 가능함을 입증하였으며, 제안된 피라미드를 이진 영상의 점진적 부호화에 이용할 경우 등가 엔트로피 측면에서 약 1.6에서 2.9배 개선될 수 있음을 보였다.

• Journal of Biosystems Engineering
• /
• 제39권4호
• /
• pp.318-323
• /
• 2014

Purpose: This study was carried out to predict the growth period and fresh weight of sprouts grown in a cultivator designed to grow sprouts under optimal conditions. Methods: The temperature, light intensity, and amount of irrigation were controlled, and images of seed sprouts were acquired to predict the days of growth and weight from pixel counts of leaf area. Broccoli, clover, and radish sprouts were selected, and each sprout was cultivated in a 90-mm-diameter Petri dish under the same cultivating conditions. An image of each sprout was taken every 24 hours from the 4th day, and the whole cultivating period was 6 days, including 3 days in the dark. Images were processed by histogram inspection, binary images, image erosion, image dilation, and the overlay image process. The RGB range and ratio of leaves were adjusted to calculate the pixel counts for leaf area. Results: The correlation coefficients between the pixel count of leaf area and the growth period of sprouts were 0.91, 0.98, and 0.97 for broccoli, clover, and radish, respectively. Further, the correlation coefficients between the pixel count of leaf area and fresh weight were 0.90 for broccoli, 0.87 for clover, and 0.95 for radish. Conclusions: On the basis of these results, we suggest that the simple image acquisition system and processing algorithm can feasibly estimate the growth period and fresh weight of seed sprouts.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2006년 학술대회 논문집 정보 및 제어부문
• /
• pp.178-180
• /
• 2006

In general, camera calibration has to be gone ahead necessarily to estimate a position and an orientation of the object exactly using a camera. Autonomous land system in order to run a vehicle autonomously needs a camera calibration method appling a camera and various road environment. Camera calibration is to prescribe the confrontation relation between third dimension space and the image plane. It means to find camera calibration parameters. Camera calibration parameters using the paved road and the unpaved road are estimated. The proposed algorithm has been detected through the image processing after obtaining the paved road and the unpaved road. There is able to detect easily edges because the road lanes exist in the raved road. Image processing method is two. One is a method on the paved road. Image is segmentalized using open, dilation, and erosion. The other is a method on the unpaved road. Edges are detected using blur and sharpening. So it has been made use of Hough transformation in order to detect the correct straight line because it has less error than least-square method. In addition to, this thesis has been used vanishing point' principle. an algorithm suggests camera calibration method using Hough transformation and vanishing point. When the algorithm was applied, the result of focal length was about 10.7[mm] and RMS errors of rotation were 0.10913 and 0.11476 ranges. these have the stabilized ranges comparatively. This shows that this algorithm can be applied to camera calibration on the raved and unpaved road.

• 경영과정보연구
• /
• 제4권
• /
• pp.257-284
• /
• 2000

According as the drivers increase who have their cars, the comprehensive studies on the automobile for the traffic safety have been raised as the important problems. Visual Recognition System for radio-controled driving is a part of the sensor processor of Unmanned Autonomous Vehicle System. When a driver drives his car on an unknown highway or general road, it produces a model from the successively inputted road traffic information. The suggested Recognition System of the Road Traffic Safety Information Board is to recognize and distinguish automatically a Road Traffic Safety Information Board as one of road traffic information. The whole processes of Recognition System of the Road Traffic Safety Information Board suggested in this study are as follows. We took the photographs of Road Traffic Safety Information Board with a digital camera in order to get an image and normalize bitmap image file with a size of $200{\times}200$ byte with Photo Shop 5.0. The existing True Color is made up the color data of sixteen million kinds. We changed it with 256 Color, because it has large capacity, and spend much time on calculating. We have practiced works of 30 times with erosion and dilation algorithm to remove unnecessary images. We drawing out original image with the Region Splitting Technique as a kind of segmentation. We made three kinds of grouping(Attention Information Board, Prohibit Information Board, and Introduction Information Board) by RYB( Red, Yellow, Blue) color segmentation. We minimized the image size of board, direction, and the influence of rounding. We also minimized the Influence according to position. and the brightness of light and darkness with Eigen Vector and Eigen Value. The data sampling this feature value appeared after building the learning Code Book Database. The suggested Recognition System of the Road Traffic Safety Information Board firstly distinguished three kinds of groups in the database of learning Code Book, and suggested in order to recognize after comparing and judging the board want to recognize within the same group with Nearest Neighborhood Decision Making.