흉부 X선 사진에 있어서 간질성 음영의 진단지원

金應珪 한밭대학교 산업과학기술연구소 1997 논문집 Vol.4 No.-

In this study, an algorithm is investigated to detect and characterize interstitial lung abnormalities seen at chest radiographs. This method includes a process of four-directional Laplacian-Gaussian filtering, and a process of linear opacity judgment. Two regions of interest(ROIs) were selected in each right lung of patients, and these ROIs were processed by our computer-analyzing system. For quantitative analysis of interstitial opacities, the radiographic index, which is the percentage of opacity areas in a ROI, was obtained and evaluated in the images. From the result, abnormal lungs were well differentiated from normal lungs. In this algorithm, the processing results were not only given as the numeric data named "radiographic index" but also confirmed with radiologists' observation on CRT. The approach, by which the interstitial abnormalities themselves are extracted, is good enough because the results can be confirmed by the observations of radiologists. In conclusion, This system is useful for the detection and characterization of interstitial lung abnormalities.

평행이동에 기인한 MRI 아티팩트의 제거

김응규 大田産業大學校 1999 한밭대학교 논문집 Vol.16 No.1

In this study, a new algorithm for canceling MRI artifact due to translational motion in image plane is described. Unlike the conventional iterative phase retrieval algorithm, in which there is no guarantee for the convergence, a direct method for estimating the motion is presented. In the previous approach, the motions in the x(read out) direction and the y(phase encoding) direction are estimated simultaneously. However, the feature of x and y directional motion is different each other. By analyzing their features, each x and y directional motion is canceled by different algorithms in two steps. First, it is noticed that the x directional motion corresponds to a shift of the x directional spectrum of the MRI signal, and the non-zero area of the spectrum just corresponds to x-axis projected area of the density function. So the motion is estimated by tracing the edges between non-zero area and zero area of the spectrum, and the x directional motion is canceled by shifting the spectrum in inverse direction. Next, the y directional motion is cancelcd by using a new constraint condition, with which the motion component and the true image component can be separated. This algorithm is shown to be effective by using a phantom image with simulated motion.

회전운동에 기인한 MRI 아티팩트의 제거

김응규 大田産業大學校 2002 한밭대학교 논문집 Vol.19 No.-

when the imaging object rotates in image plane during MRI scan. its rotation cause phase error and non-uniform sampling to MRI signal. Therefore the following methods is presented to improve the quality of the MR image which includes the artifact. The first, an algorithm to correct the k space non-uniform sampling data is considered by using bilinear interpolation and super-position property with the given rotational parameters. The second, assuming that the angle of 2-D rotational motion is already known and the position of 2-D rotational center is unknown, an algorithm to correct the artifact which is based on the phase correction is considered. The third, in case of 2-D rotational motion with unknown rotational center and unknown rotational angle, an algorithm is presented to correct the MRI artifact. At this case, the energy of an ideal MR image is minimum outside the boundary of the imaging object to estimate unknown motion parameters and the measured energy increase when the imaging object has an rotation. By using this property, an evaluation function is defined to estimate unknown values of rotation angle at each phase encoding step. Finally, the effectiveness of this presented techniques is shown by using a phantom image with simulated motion and a real image with 2-D translational shift and rotation.

주사형 정전용량 현미경의 개발

金應珪,權寧島 大田産業大學校 1996 한밭대학교 논문집 Vol.13 No.0201

In this study, a scanning capacitance microscopy(SCaM) by stage driving is proposed and presented some of the experimental results. SCaM is a microscopy which scans a surface of materials mechanically in two or two point five dimensions by a capacitance probe with a few tenth ㎛ size tip, and display images of the surface shape or capacitive distribution. The present target of the SCaM is 0.1㎛ resolution power which exceeds that of optical microscopy. This will become a powerful tool for inspecting ULSI pattern etched by X-ray, biological data, etc. The experimental system is composed based on a VHD video disk which captures the capacitance changes of the video disk surface and converts it into video signal.

웨이블릿 변환에 의한 정보 압축

김응규,권영도 한밭대학교 정보통신전문대학원 2005 정보통신전문대학원 논문집 Vol.3 No.1

본 연구에서는 웨이블릿 변환법에 기반한 정보압축을 제안한다. 신호와 화상의 압축 원리는 그 전력이 낮은 영역쪽으로 집중한다는 것을 이용한 비트의 할당, 즉 최적 양자화로, 주파수 분해방법으로써 이산코사인 변환법과 하부밴드 부호화 등이 실용화되고 있다. 본 연구에서 취급하는 웨이블릿 변환법도 그중의 한가지로 신호를 다중해상도로 분해해서 최적양자화를 행한다. 웨이블릿 변환은 주파수분할 방법 중의 하나로서 이산코사인 변환 또는 서브밴드 필터링 그리고 필터뱅크로 구현할 수 있다. 또한 공간적으로 국한(집중)되는 기저함수의 우수한 성질로부터 다른 방법에서 지적되어진 단점을 억제하는 효과가 있어, 다른 방법과 비교하면서 웨이블릿 변환의 장점에 대해서도 기술한다. In this study, on information compression based on the wavelet technique is described. The principle of signal or image compression is performed by optimization of quantization, that is the bit allocation taking advantage of their energy concentration in low frequency components. The wavelet transform is one of frequency decomposition, such as the discrete cosine transform or sub-band filtering, and it is also implemented as a filter bank. Wavelet transform with use of spatially localized basis function can reduce several drawbacks in conventional methods. The benefit of wavelet based compression method is described as comparing the transform method to another ones.

신경회로망을 이용한 다소자 초음파 트랜스듀서에 의한 수중물체의 화상화

金應珪 大田産業大學校 1996 한밭대학교 논문집 Vol.13 No.1

Comparing with light, ultasound has so small degenerating characteristic that measurement using ultrasound is widely used in medicine and biology. In this paper, a multi-element ultrasound tanducer has been developed aiming at basic experiment of 3D endovascular ultrasound endoscopy for clinical diagnosis. A three-layer artilficial neural network is employed for learning and imaging of targets placed in front of the transducer. Inputs to the neural network are ultrasound echos received by transducer elements. Back propagation is used as a learning algoritghm. Rough visualization is realized even for objects with unlearned shapes after being trained by primitive patterns of various sized rectangular targets.

레이져스캐너 3차원계측에 있어서 다방향 데이터의 통합

金應珪,權寧島 大田産業大學校 1995 한밭대학교 논문집 Vol.12 No.0201

When objects are scanned satially by a laser-beam and mechanical mirror scanners, a technique is presentd for obtainment of three-dimensional data, and then measurement errors are often a significant problem. In this paper, first, a model of the systematic errors for a two-dimensional mirror scanner is developed based on the geometry of the laser scanning system. Parameter values are modified to reduce the different between measured and calculated values. This show that it is possible to measure three-dimensional data in a three-dimensional measurement system with enough precision for practical use. Futhermore, complete object data can be obtained without occlusion by combine scanner data from multiple directions.

흉부 X선 사진에 있어서 폐의 간질성 음영의 컴퓨터 해석

김응규 大田産業大學校 1998 한밭대학교 논문집 Vol.15 No.1

In this study, a computer analysis method is presented to detect and quantify interstitial lung shadows in chest radiographs. This method includes a process of four-directional Gaussian-Laplacian filtering, and a process of determination and subtraction of line, each right lung of patients, and these regions of interest are processed by computer analysis system. For detection and quantification of interstitial shadows of lung diseases, three kinds of parameters, that is, the radiographic index(D_bin) obtained from binarization, the index(D_lin) obtained from linear shadow determination and the index(D_sub) obtained from linear shadow subtraction of processed images, and the other parameter, that is, the combined radiographic index(D_com) obtained from D_lin and D_sub, are calculated and evaluated. In ROC(Receiver Operating Characteristic) analysis, the value of AZ(Area under the ROC curves) of each parameter was greater than 0.9, indicating excellent detectability of interstitial lung diseases. Especially, among the parameters, D_com showed the best performance in ROC analysis. These indices shows that abnormal lungs are well differentiated from normal lungs, and also are confirmed with radiologists' observation on CRT. This computer analysis method including radiographic indices is thought to be useful for the detection and quantification of interstitial lung shadows.

음성 특징량과 입모양 화상을 병용하는 신경회로망을 이용한 모음인식

金應珪 大田産業大學校 1996 한밭대학교 논문집 Vol.13 No.1

This paper describes on a neural approach intending to imfrove performance of a voice recognition device by using not only voice features but also image features of the mouth shape. FFT power spectrum is used as the voice features and gray level image, binary image, and its geometrical shape features are effective for the voice recognition by a neural network Vowel recognition rate of about 80% is obtained by using only voice features, and 91% by using voice features plus binary image for unrestricted speaker. This method can be applied not only the improvement of the voice recognition but also communication-aid for hearing impaired people.

국소영역의 직선성을 고려한 가중치부가 하아프 변환에 의한 직선검출

金應珪 大田産業大學校 1995 한밭대학교 논문집 Vol.12 No.0201

The problem of detecting lines in a digital image is one of the most fundamental problem in pattern recognition. The Hough Transform(HT) is well-known techniqe for its effectivity to detect straight lines from a noisy image. The standard HT consists of two stages, the first stage is voting and the second one is the peak detection in the line parameter space. The voting is performed in the (ρ, θ) line parameter space with the same weight for all θ , and the space is quantized to accumulate the voting from a digital image. To get more reasonable result of line detection, The linearity in local region is introduced, and a method of voting the linearity into parameter space is proposed. The experiments show that the voting method gives the robust result of line detection even if a given image has characters as well as lines.