Choonsuk Oh

Development of a noncontact optical sensor for measuring the shape of a surface and thickness of transparent objects

This paper deals with the non-contact optical sensor to measure the shape and the thickness of transparent plates such as glass panel of LCD(Liquid Crystal Display). The conventional methods to obtain the shape or thickness of transparent plates are contact type sensor such as LVDT(Linear Variable Differential Transformer). Due to the contact between the tip of the sensor and the surface of objects, the tip is abraded. In addition the casting glass under high temperature results in extendin g the size of sensor body. The accuracy of the sensor is degraded due to these reasons. In this paper, to overcome these problems, we proposed a low cost non-contact optical sensor that is composed of a Hologram laser unit used for optical pickup of CD player and a plastic lens. To evaluate the performance of the proposed optical sensor, a series of experiments ware performed for various measurement conditions. Based upon observation of the experimental results, the proposed sensor shows a good performance in measuring the shape of transparent plates.

The Study of Improving the Accuracy in the 3D Data Acquisition of Motion Capture System

We introduced the motion capture system using two CCD cameras recently, but could not show any better accuracy than a system using PSD camera. In this paper, we propose two kinds of method to improve the accuracy of the 3D acquisition of the motion capture system using CCD cameras. The applied methods are a distortion removal and z-axis adjustment. We show the result how much the accuracy on the 3D acquisition system improved through comparing with the previous system.

Video tape recorder head inspection using image processing techniques

In currently selling video tape recorders (VTRs), two or more heads fixed at a revolving drum are used to record or reproduce video signals. To obtain uniform quality, the video heads should have similar characteristics such as geometrical shape and electromagnetic performance. This research is designed to develop a visual inspection system that can measure four geometrical parameters. To obtain reliable results, a digital signal processing-based image board and target-based imageprocessing algorithms were used. A series of experiments were performed on 1000 video heads. From observation of the experimental results, except for the cases of noise on the head and cracklike defects, we were able to obtain high inspection reliability, with an error bound within ±5 ?m.

Off-line recognition of handwritten Korean and alphanumeric characters using hidden Markov models

This paper proposes a recognition system of constrained handwritten Hangul (Korean characters) and alphanumeric characters using discrete hidden Markov models (HMM). HMM process encodes the distortion and similarity among patterns of a class through a doubly stochastic approach. Characterizing the statistical properties of characters using selected features, a recognition system can be implemented by absorbing possible variations in the form. Hangul shapes are classified into six types, and their recognition based on quantized features is performed by optimally ordering features according to their effectiveness in each class. The constrained alphanumerics recognition is also performed using the same features employed in Hangul recognition. The forward-backward, Viterbi, and Baum-Welch reestimation algorithms are used for training and recognition of handwritten Hangul and alphanumeric characters. Simulation results show that the proposed method recognizes handwritten Hangul and alphanumerics effectively.

Study on the center of rotation method based on minimum spanning tree matching algorithm for fingerprint recognition

The minimum spanning tree (MST) matching algorithm has been used for searching the partial matching points extracted from fingerprint images. The method, however, has some limitations. To obtain the relationship between the enrolled fingerprints and the input fingerprints the MST is used to generate the tree graph that represents the unique graph for the given minutiae. From the graph, the matching points are estimated. However, the shape of the graph highly depends on the positions of the minutiae. If there are some pseudo minutiae caused by noise, the shape of the graph will be totally different. To overcome the limitations of the MST, we propose the center of rotation method (CRM) that finds true partial matching points. The proposed method is based on the assumption that the input fingerprint minutiae are rigid body. In addition, we employ the polygon wrapping the minutiae for fast matching speed and reliable matching results. In conclusion we have been able to confirm fast performance and high identification ratio by using CRM.

A study on enhanced algorithms for detecting defects of glasses

In this paper, we proposed image processing algorithms and developed the vision system for measuring blisters of flat glasses. The images captured by changing focus of camera because the depth of field of CCD camera is smaller than the depth of glasses and classified two types in blister as circle and ellipse types by the proposed criteria of measurement. We applied different methods to each type to improve the accuracy. In the pre-processing, it obtained binarized images to find blister. The information such as area, boundary, length, and position of blister has extracted from the binarized images by using chain code algorithm. For an illumination, the bright field illumination as back-light was applied. We tested the automatic measurement system, which we developed by using vision system, It successes 100% for detecting the blister at all glasses, and the error accuracy of information is about +/-10%.

RF Signal Synchronized Low Cost Motion Capture Device

This paper describes RF synchronized monocular PSD-based motion capture sensor to employ with commercial video game systems such as Microsofts XBOX and Sonys Playstation II. The system is compact, low-cost, and only requires a one-time calibration at the factory. The system includes a PSD(Position Sensitive Detector), active infrared (IR) LED markers that are placed on the object to be tracked and RF module to synchronize the marker LED with PSD MCU. A series of experiments were performed to evaluate the performance of our prototype system. From the experimental results we see that the proposed system has the advantages of the compact size, the low cost, the easy installation, and the high frame rates to be suitable for high speed motion tracking in games.

Implementation of a 3D Recognition applying Depth map and HMM

Recently, we used to recognize for human motions with some recognition algorithms. examples, HMM, DTW, PCA etc. In many human motions, we concentrated our research on recognizing fighting motions. In previous work, to obtain the fighting motion data, we used motion capture system which is developed with some active markers and infrared rays cameras and 3 dimension information converting algorithms by the stereo matching method. In this paper, we describe that the different method to acquiring 3 dimension fighting motion data and a HMM algorithm to recognize the data. One of the obtaining 3d data we used is depth map algorithm which is calculated by a stereo method. We test the 3d acquiring and the motion recognition system, and show the results of accuracy and performance results.

Comparing two methods for acquiring 3D data of motion capture system by using PSD camera and CCD camera

This paper presents a monocular PSD-based motion capture system and CCD camera based system. The first system includes a PSD (Position Sensitive Detector) and active infrared (IR) LED markers that are placed on the object to be tracked. The PSD sensor is placed in the focal plane of a wide-angle lens. The microcontroller calculates the 3D position of the markers using only the measured intensity and the 2D position on the PSD. The second system includes two CCD camera and LED markers that are in hand and draw any motion trace. From the experimental results we see that the two kinds of methods for acquiring 3d data will be compare in accuracy, speed, process, methods, experimental circumstance.

An Implementation of OPC Client for Display Inspection System

OPC is the preferred communication standard for sharing process control data at all levels of the enterprise and becomes the global standard, sends/receives the data with PLC programs of inspection system via network, and manages various data using OPC server. In this paper, we introduce the developed OPC client system for FPD inspection system with OPC server. To control inspection system through OPC server, we have developed libraries which consist of stage driver, switch driver, microscope driver, image processing, alignment, and task management modules. Also we have developed the test program to verify all functions precisely. We describe the mechanism of inspection system, OPC connection method, and programs.