Earticle

본 논문은 기울기가 5도 이상 기울어진 자동차 번호판 영상의 숫자 및 한글 인식률을 높이고자 번 호판 영상을 회전시키려 한다. 먼저, 입력 영상은 자동차 번호판 RGB 색상의 영상으로 이 RGB 영상 을 흰색 계열로 보정한 뒤 Gray scale 영상으로 바꾼다. 바뀐 Gray scale 영상을 이용하여 배경과 글 자색은 0(검정색)으로 지정하며 번호판 색상은 255(하얀색)로 지정한 뒤 영상을 이진화 한다. 이진화 된 영상의 영상 회전 방향에 따라 해당하는 꼭짓점의 좌표를 구한다. 마지막으로 아크탄젠트를 이용 하여 구해진 꼭짓점의 좌표로 기울어진 각도를 구한 뒤 회전 소스를 이용하여 각도만큼 번호판 영상 을 회전시킨다. 회전 된 번호판 영상은 각종 문자 인식이나 숫자 인식에 이용되며 회전 알고리즘을 이용하면 이용하지 않았을 때보다 문자 혹은 숫자에 대한 인식률이 높아짐을 확인하였다.

To increase the recognition rate of numbers and Hangul characters in tilted car license plates images, this paper suggests to rotate the images tilted more than 5 degrees. First, the input RGB image is revised by the white image and converted to the Gray scale image. After setting the background and text colors as 0(black) and the color of plates as 255(white) with this Gray scale image, the image is binary coded. And then, the coordinates of the vertices for each rotation direction of the binary coded image are obtained. Finally, using arc tangent, this rotation algorithm calculates the slanting angle with the obtained coordinates of the vertices and rotates the licence plate image with rotate sauce. This Rotation algorithm can be adopted to various character or number recognitions and give higher recognition rate than existing recognition system.

Feature-points matching is an important concept in binocular stereo vision. The procession of multi-scale feature-points matching in classical Harris-SIFT algorithm is time-consuming and has high complexity when describing the feature-points. This paper proposed a new improved Harris-SIFT algorithm based on rotation-invariant LBP (Local binary patterns) operator. Firstly, the Harris operator is used to extract feature points from DOG (Difference of Gaussian) scale space. Then, the dominant direction of feature point is calculated and 81-dimensional rotation-invariant LBP descriptors are extracted when the rotation matching window is coordinated to this direction. At last, Best-Bin-First (BBF) algorithm is used to search the matching points between the two sets of feature points. Experimental results show that the proposed algorithm is lower time-consuming than classical Harris-SIFT algorithm and remains the similar matching correct rate.

영상 회전은 영상 처리나 영상 패턴 인식에서 중요한 전처리 방법 중 하나이다. 영상 회전은 회전 행렬의 곱으로 이루어 진다. 그러나 기존의 방법은 대량의 실수 연산과 삼각 함수 계산을 필요로 하므로 수행 시간이 오래 걸린다. 본 논문에서는 이 같은 두가지 주요 지체 연산과정을 제거한 새로운 고속 영상 회전 알고리즘을 제안한다. 제안된 알고리즘은 단지 2개의 전단 연산을 행하므로 매우 빠르다. 또한 최신 병렬 처리 기술인 CUDA를 적용한다. CUDA는 최근 널리 보급된 GPU를 이용한 대용량 병렬처리 계산 아키텍쳐이다. GPGPU는 그래픽 전용프로세서이므로 화소 단위의 병렬처리에 탁월한 성능을 보인다. 제안된 알고리즘은 기존의 회전 알고리즘과 다양한 크기의 영상에 대해 비교 실험한다. 실험 결과는 제안된 알고리즘이 기존의 방법보다 8배 이상의 매우 우수한 성능을 보인다.

Image rotation is one of main pre-processing step in image processing or image pattern recognition. It is implemented with rotation matrix multiplication. However it requires lots of floating point arithmetic operations and trigonometric function calculations, so it takes long execution time. We propose a new high speed image rotation algorithm without two major time-consuming operations. It use just 2 shear translation operations, so it is very fast. In addition, we apply a parallel computing technique with CUDA. CUDA is a massively parallel computing architecture using prevailed GPU recently. As GPU is a dedicated graphic processor, it is exellent for parallel processing of pixels. We compare the proposed algorithm with the conventional rotation one with various size images. Experimental results show that the proposed algorithm is superior to the conventional rotation ones.

이미지 인식 및 영상처리, 컴퓨터 비전 등의 분야에서 합성곱 인공신경망 (Convolutional Neural Network, CNN)은 다양하게 응용되고 탁월한 성능을 내고 있다. 본 논문에서는 CNN을 활용한 이미지 인식 시스템에서 인식률을 저하시키는 요인 중 하나인 이미지의 회전에 대한 해결책으로써 CNN 기반 이미지 회전 보정 알고리즘을 제안한다. 본 논문에서는 Leeds Sports Pose 데이터셋을 활용하여 이미지를 임의의 각도만큼 회전시킨 학습데이터로 인공지능 모델 을 학습시켜 출력으로 회전된 각도를 추정하도록 실험을 진행하였다. 학습된 인공지능 모델을 100장의 테스트 데이터 이미지로 실험하여 mean absolute error (MAE) 성능지표를 기준으로 4.5951의 값을 얻었다.

Recently, convolutional neural network (CNN) have been showed outstanding performance in the field of image recognition, image processing and computer vision, etc. In this paper, we propose a CNN-based image rotation correction algorithm as a solution to image rotation problem, which is one of the factors that reduce the recognition rate in image recognition system using CNN. In this paper, we trained our deep learning model with Leeds Sports Pose dataset to extract the information of the rotated angle, which is randomly set in specific range. The trained model is evaluated with mean absolute error (MAE) value over 100 test data images, and it is obtained 4.5951.

We propose a fast algorithm using Extreme Contour Point (ECP) to detect the angle of rotated images, is implemented by rotation feature of one covered frame image that can be applied to correct the rotated images like in image processing for real time applications, while CORDIC is inefficient to calculate various points like high definition image since it is only possible to detect rotated angle between one point and the other point. The two advantages of this algorithm, namely compatibility to images in preprocessing by using Sobel edge process for pattern recognition. While the other one is its simplicity for rotated angle detection with cyclic shift of two $1{\times}n$ matrix set without complexity in calculation compared with CORDIC algorithm. In ECP, the edge features of the sample image of gray scale were determined using the Sobel Edge Process. Then, it was subjected to binary code conversion of 0 or 1 with circular boundary to constitute the rotation in invariant conditions. The results were extracted to extreme points of the binary image. Its components expressed not just only the features of angle ${\theta}$ but also the square of radius $r^2$ from the origin of the image. The detected angle of this algorithm is limited only to an angle below 10 degrees but it is appropriate for real time application because it can process a 200 degree with an assumption 20 frames per second. ECP algorithm has an O ($n^2$) in Big O notation that improves the execution time about 7 times the performance if CORDIC algorithm is used.

There are two rotation matrix parameters in a model, pro-posed by Prentice in 1989, for pairs of rotations in 3 dimensional space. For the least squares estimates of the two parameters, an algorithm was also presented, but it turned out that the algorithm could fail to get the least squares estimates. This article provides another algorithm for the least squares estimates and its performance is demonstrated by simulation results.

As a lot of researches on the speech signal processing are performed due to the recent rapid development of the information-communication technology. the pitch period is used as an important element to various speech signal application fields such as the speech recognition. speaker identification. speech analysis. or speech synthesis. A variety of algorithms for the time and the frequency domains related with such pitch period detection have been suggested. One of the pitch detection algorithms for the time domain. AMDF (average magnitude difference function) uses distance between two valley points as the calculated pitch period. However, it has a problem that the algorithm becomes complex in selecting the valley points for the pitch period detection. Therefore, in this paper we proposed the modified AMDF(M-AMDF) algorithm which recognizes the entire minimum valley points as the pitch period of the speech signal by using the rotation transform of AMDF. In addition, a threshold is set to the beginning portion of speech so that it can be used as the selection criteria for the pitch period. Moreover the proposed algorithm is compared with the conventional ones by means of the simulation, and presents better properties than others.

A novel Genetic Algorithm(GA)-based method is suggested to transform a coupling matrix to another, without the procedure of Matrix Rotation. This can remove tedious work like pivoting and deciding rotation angles needed for each of the iterations. The error function for the GA is simply formed and used as part of error minimization for obtaining the solution. An 8th order dual-mode elliptic integral function response filter is taken as an example to validate the present method.

Conventional inertial measurement units cannot be used in the spinning vehicle with high rotation rate due to gyro's narrow operation range. By the way, angular acceleration can be measured using the accelerometer array distributed in the vehicle. This paper derives a mechanization for the gyro-free INS in the navigation frame, and proposes a gyro-free INS algorithm based on the derived mechanization. In addition, the proposed algorithm is used to estimate angular velocity, attitude, velocity, and position of a spinning vehicle with high rotation rate. A MATLAB-based software platform is configured in order to show validation of the proposed algorithm. The reference trajectory of a spinning vehicle at 3 round per second, 30 round per second are set up, and the outputs of accelerometer are generated when triads of accelerometer are located at the origin and all the axes. Navigation results of the proposed algorithm for the generated output are presented. The results show that the proposed navigation algorithm can be applied to the spinning vehicle with high rotation rate.

The purpose of this study is to develop a practical image inspection system that could recognize it correctly, endowing flexibility to the productive field, although the same object for work will be changed in the size and rotated. In this experiment, it selected a fighter, rotating the direction from $30^{\circ}\;to\;45^{\circ}$ simultaneously while changing the size from 1/4 to 1/16, as an object inspection without using another hardware for exclusive image processing. The invariant moments, Hu has suggested, was used as feature vector moment descriptor. As a result of the experiment the image inspection system developed from this research was operated in real-time regardless of the chance of size and rotation for the object inspection, and it maintained the correspondent rates steadily above from 94% to 96%. Accordingly, it is considered as the flexibility can be considerably endowed to the factory automation when the image inspection system developed from this research is applied to the productive field.

최근, 다양한 형태의 웨어러블 컴퓨팅 디바이스와 이에 따른 응용 프로그램이 개발되고 있으며, 이들 중에서 헬스 케어의 한 영역으로 웨어러블 컴퓨팅 디바이스를 이용하여 개인의 운동량은 측정하는 다양한 연구가 진행되고 있다. 본 논문에서는 가장 기초적인 운동인 걸음걸이 측정 알고리즘으로 중력 제거 및 백터 회전(Gravity Removal and Vector Rotation) 알고리즘을 제안하고, 이를 위한 실험으로 보행 또는 주행 중인 개인의 다양한 위치에 부착한 웨어러블 디바이스에 장착된 3축 MEMS 가속도 센서로부터 획득된 가속도 값을GRVR 알고리즘을 이용하여 걸음 수를 측정한다. GRVR은 가속도 센서로 획득된 3축 가속도 값으로 부터 중력에 의한 가속도 분은 저대역 필터를 이용하여 제거 하고, 이 후 각각의 가속도 값에서 직류 성분 제거 및 센서 축 회전 보상인 GRVR 알고리즘으로 보행 혹은 주행 중에 순수하게 걸음걸이에 의하여 발생하는 가속도 변화분 만을 추출한다. 실험 결과로 웨어러블 디바이스를 개인의 허리 중앙 혹은 우측에 부착한 경우 GRVR 알고리즘을 이용한 걸음 수 측정은 99.4%의 정확도, 또한 손목에 부착한 경우 상용 3축 가속도 만보계의 83%보다 정확한 91.1%의 정확도를 확인하였다.

In this paper, we propose Gravity Removal and Vector Rotation algorithm for counting steps of wearable device, and we evaluated the proposed GRVR algorithm with Micro-Electro-Mechanical (MEMS) 3-axis accelerometer equipped in low-power wearable device while the device is mounted on various positions of a walking or running person. By applying low-pass filter, the gravity elements are canceled from acceleration on each axis of yaw, pitch and roll. In addition to DC-bias removal and the low-pass filtering, the proposed GRVR calculates acceleration only on the yaw-axis while a person is walking or running thus we count the step even if the wearable device's axis are rotated during walking or running. The experimental result shows 99.4% accuracies for the cases where the wearable device is mounted in the middle and on the right of the belt, and 91.1% accuracy which is more accurate than 83% of commercial 3-axis pedometer when worn on wrist for the case of axis-rotation.

In this paper we proposed an algorithm for rotation-invariant pattern recognition and rotated angle estimation between two patterns by employing selective template matching. Generally template matching has been used in determining the location of pattern but template matching requires a number of calculating correlation. To reduce the number of correlation we used steady-state genetic algorithm which is effective in optimization problem. We apply this method to distinguish specific pattern from similar coin patterns and estimate rotated angle between patterns. Our result leads us to the conclusion that proposed method performed faster than classical template matching

본 논문에서는 지문 영상의 개별적이고 부분적인 특성(밝기, 품질 차이 등)을 고려하여 지문의 특징량 추출을 수행함으로써 지문 입력기간의 차이에서 오는 문제점을 극복하며, 정합 과정시에는 회전 및 크기 변화, 위치 이동 등에 관계없이 균일한 성능을 보장하면서도 실시간 내에 처리가 가능한 자동 지문인식 알고리즘을 제안한다. 특히 본 방법에서는 지문의 분류 작업을 융선의 전체적인 흐름 모양과 중심점, 삼각주와 같은 특이점간의 상호 관계를 이용하여 결정하게 된다. 개발된 알고리즘을 평가하기 위해 NIST 지문 및 한국인의 연령과 직업별 인구 분포에 따라 구축된 압착 날인 지문, 그리고 회전 날인 후 수치화된 지문과 같은 다양한 영상들에 대해 제안된 방법을 적용하였으며, 전체적인 인식 성능이 타인 수락 오류율을 0%로 설정했을 때 본인 거부 오류율이 0.07%라는 결과를 얻을 수 있었다.

In this paper, an algorithm for a real-time automatic fingerprint identification system is proposed. The fingerprint feature volume is extracted by considering distinct and local characteristics(such as intensity and image quality difference etc.) in fingerprint images, which makes the algorithm properly adaptive to various image acquisitionj methods. Also the matching technique is designed to be invariant on rotation, scaling and translation (RST) changes while being capable of real-time processing. And the classification of fingerprints is performed based on the ridge flow and the relations among singular points such as cores and deltas. The developed fingerprint identification algorithm has been applied to various sets of fingerprint images such as one from NIST(National Institute of Standards and Technology, USA), a pressed fingerprint database constructed according to Korean population distributions in sex, ages and jobs, and a set of rolled-than-scanned fingerprint images. The overall performance of the algorithm has been analyzed and evaluated to the false rejection ratio of 0.07% while holding the false acceptance ratio of 0%.

In this paper, we studied for the real-time azimuthal measurement of HMD(Head Mounted Display) using the feature points detection to control the tele-operated vision system on the mobile robot. To give the sense of presence to the tele-operator, we used a HMD to display the remote scene, measured the rotation angle of the HMD on a real time basis, and transmitted the measured rotation angles to the mobile robot controller to synchronize the pan-tilt angles of remote camera with the HMD. In this paper, we suggest an algorithm for the real-time estimation of the HMD rotation angles using feature points extraction from pc-camera image.

This paper proposes an enhanced algorithm for person identification system utilizing hand vein pattern. The conventional algorithm does not cope with distortion caused by image rotation caused by misplaced hands on the imaging device. A straightforward approach to consider the rotaional compensation required too much computational load, thus, we devised an approach to expect the rotation direction along with image translation, reducing the compuational requirement dramatically In this paper, we present the details of the algorithm with experimental results with the new algorithm.

Recently, due to the rapid development of multimedia technologies, as image data has been extensive and large-scaled, the problem of increasing the time needed to retrieve the desired image is gradually critical. Image retrieval system that allows users to quickly and accurately search for the desired image information has been researched for a long time. However, in the case of content-based image retrieval representative Color Histogram, Color Coherence Vectors (CCV), Scale Invariant Feature Transform (SIFT) used in sensitive to changes in brightness, rotation, there is a problem that can occur misrecognized division off the power. In this paper, in order to evaluate the video retrieval system proposed, no change in brightness, respectively 0°, 90°, 180°, 270° rotated brightness up based on the case of changing, when the brightness down the results were compared with the performance evaluation of the system is an average of about 2% to provide the difference in performance due to changes in brightness, color histogram is an average of about 12.5%, CCV is an average of about 12.25%, it appeared in the SIFT is an average of about 8.5%, Thus, the proposed system of the variation width of the smallest in average about 2%, was confirmed to be robust to changes in the brightness and rotation than the existing systems.

내접형 기어펌프에 사용되는 지로터의 경우, 소결 공정기술의 발달로 복잡한 형상의 가공이 용이하며 내구성과 소형화의 장점이 있으나, 자동차의 연비효율 개선을 위한 유량증대 및 소음저감을 위한 연구가 요구되고 있다. 본 논문에서는 기존의 병진이동에 의한 지로터 치형의 형상조합 알고리즘이 가지는 형상 불연속성을 개선하기 위하여 회전이동 및 병진이동을 이용한 형상조합 알고리즘을 개발하였으며, 이를 토대로 개발된 설계 프로그램을 이용하여 3-타원이 조합된 경우와 타원 1-인벌루트-타원 2 이조합된 경우에 대해 치형형상을 도출하고, 형상에 따른 로터의 주요성능(유량, 유량맥동, 미끄럼률 및 압력각)을 계산하였다. 해석결과를 토대로 조합된 치형형상에 대한 최적설계를 수행하였으며, 기존 치형과의 성능 비교를 통하여 설계의 타당성을 검증하였다.

Because technology of the sintering process is highly advanced, the gerotor used in the internal gear pumps of vehicles has advantages for manufacturing complex profiles and obtaining durability and minimization. However, it has been necessary to continuously improve the flow rate and noise of internal gear pumps for better fuel efficiency. The existing rotor was designed using a translation algorithm. This caused a discontinuity of the rotor profile, which had adverse effects on the performance. In this study, to improve the discontinuity of the profile, a new design program using a rotation and translation algorithm was developed, and two types of combined multiple profiles (three-ellipses and ellipse 1-involute-ellipse 2) were generated. Then, the performances (flow, flow rate, specific sliding, and pressure angle) of these profiles were calculated. On the basis of the calculation results for the performances, optimal designs of the two types were carried out and verified by comparing their performances with those of the existing rotor profiles.

A problem of relevant interest to some industries is that of the optimum two-dimensional layout. In this problem, one is given a number of rectangular sheets and an order for a specified number of each of certain types of two-dimensional regular and irregular shapes. The aim is to cut the shapes out of the sheets in such a way as to minimize the amount of waste produced. In this paper, we propose a genetic algorithms using rotation parameters by which the best pattern of layout is found.