• 전자공학회논문지S
• /
• 제36S권11호
• /
• pp.95-101
• /
• 1999

본 논문은 연속 영상에서 윤곽선과 특징을 이용하여 주위 환경 변화에 적응가능한 다중 물체 추적 방법을 제안한다. 적응 배경 모델을 사용하여 주위 환경 변화에 적응케 했으며, 물체 분할 모델은 배경 영상과 현재 영상의 차영상에서 국부 영상의 임계값 이상의 화소를 찾아 연결한 영역을 추출한다. 특징 추출과 물체인식모델은 탐색 창 내에서 발견된 다중 물체의 데이터 연상 문제를 해결하기 우해 사용되며, 실시간 추적을 위해 칼만 필터를 사용하였다. 제안된 방법을 도로 영상에 적용한 결과 다중 차량 추적이 정확히 이루어짐을 실험을 통해 보였다.

• 한국통신학회논문지
• /
• 제38A권2호
• /
• pp.174-182
• /
• 2013

Most methods for fusion-based finger vein recognition were to fuse different features or matching scores from more than one trait to improve performance. To overcome the shortcomings of "the curse of dimensionality" and additional running time in feature extraction, in this paper, we propose a finger vein recognition technology based on matching score-level fusion of a single trait. To enhance the quality of finger vein image, the contrast-limited adaptive histogram equalization (CLAHE) method is utilized and it improves the local contrast of normalized image after ROI detection. Gabor features are then extracted from eight channels based on a bank of Gabor filters. Instead of using the features for the recognition directly, we analyze the contributions of Gabor feature from each channel and apply a weighted matching score-level fusion rule to get the final matching score, which will be used for the last recognition. Experimental results demonstrate the CLAHE method is effective to enhance the finger vein image quality and the proposed matching score-level fusion shows better recognition performance.

• 대한임베디드공학회논문지
• /
• 제10권5호
• /
• pp.325-333
• /
• 2015

Vision based night-time vehicle detection has been an emerging research field in various advanced driver assistance systems(ADAS) and automotive vehicle as well as automatic head-lamp control. In this paper, we propose night-time vehicle detection method based on multi-class support vector machine(SVM) that consists of thresholding, labeling, feature extraction, and multi-class SVM. Vehicle light candidate blobs are extracted by local mean based thresholding following by labeling process. Seven geometric and stochastic features are extracted from each candidate through the feature extraction step. Each candidate blob is classified into vehicle light or not by multi-class SVM. Four different multi-class SVM including one-against-all(OAA), one-against-one(OAO), top-down tree structured and bottom-up tree structured SVM classifiers are implemented and evaluated in terms of vehicle detection performances. Through the simulations tested on road video sequences, we prove that top-down tree structured and bottom-up tree structured SVM have relatively better performances than the others.

• 한국멀티미디어학회논문지
• /
• 제3권3호
• /
• pp.253-262
• /
• 2000

본 논문에서는 적응적으로 추출된 입력 데이터의 특징을 은닉층 뉴런 개수와 중심값 설정에 이용하는 새로운 radial basis 함수 신경망을 제안하였다. 제안된 신경망에서는 입력데이터의 특징을 효과적으로 추출하기 위해 적응 학습알고리즘의 주요성분분석 기법을 이용하였다. 이렇게 하면 주요성분분석 기법이 가지는 대용량의 입력데이터를 통계적으로 독립인 특징들의 집합으로 변환시키는 장점과 RBF신경망이 가지는 우수한 속성을 그대로 살릴 수 있다. 제안된 기법의 radial basis 함수 신경망을 200명의 암환자를 2부류(초기와 악성)로 분류하는 문제에 적용하여 시뮬레이션한 결과, k-평균 군집화 알고리즘을 이용한 radial basis 함수 신경망에 의한 결과와 비교할 때 학습시간과 시험 데이터의 분류에서 더욱 우수한 성능이 있음을 확인할 수 있었다. 그리고 신경망의 초기 연 결가중치에 대한 의존도와 평활요소의 설정여유도 측면에서도 우수한 특성이 있음을 확인할 수 있었다.

• International journal of advanced smart convergence
• /
• 제4권2호
• /
• pp.20-28
• /
• 2015

In this paper, we propose an effective tracking algorithm with an appearance model based on features extracted from a video frame with posture variation and camera view point adaptation by employing the non-adaptive random projections that preserve the structure of the image feature space of objects. The existing online tracking algorithms update models with features from recent video frames and the numerous issues remain to be addressed despite on the improvement in tracking. The data-dependent adaptive appearance models often encounter the drift problems because the online algorithms does not get the required amount of data for online learning. So, we propose an effective tracking algorithm with an appearance model based on features extracted from a video frame.

• 한국정보전자통신기술학회논문지
• /
• 제12권3호
• /
• pp.179-185
• /
• 2019

본 논문은 저전력 특징추출 알고리즘의 구현을 위한 블록 유형 분류 기반 영상 이진화 방법을 제안한다. 제안하는 방법은 영상 내에서 $64{\times}64$ macro block 크기로 영상을 나누고 각 블록 유형별 threshold 값을 한 번만 연산한 후 그 값을 re-use 하는 기법으로 구현될 수 있다. 알고리즘은 threshold 값이 같은 영상/블록 유형 내에서 최대 9%의 변화율만 발생하는 것을 정량적인 결과를 기반으로 검증했다. 기존 알고리즘은 $512{\times}512$ 이미지 기준으로 macro block을 $64{\times}64$로 나누었을 때 64개의 블록을 위해 threshold 값을 연산해야 하지만 제안하는 방법은 모두 같은 블록 유형이 출력되는 best case의 경우 threshold 연산을 한번만 수행하고, 나머지 63개의 블록에 대해서는 블록 유형 구분 과정만 수행하면 adaptive threshold calculation 연산을 98% 생략할 수 있다. 모든 블록 유형이 발생하는 worst case일 때 threshold calculation 연산은 다섯 번 수행되고, 나머지 59개의 블록에 대해서는 블록 유형 구분 과정만 수행할 수 있으므로 93%의 adaptive threshold calculation 연산을 생략할 수 있다.

• 한국산업융합학회 논문집
• /
• 제5권1호
• /
• pp.33-38
• /
• 2002

This paper proposes an efficient method for improving the performance of polynomial adaline using the dimension reduction of independent variables. The adaptive principal component analysis is applied for reducing the dimension by extracting efficiently the features of the given independent variables. It can be solved the problems due to high dimensional input data in the polynomial adaline that the principal component analysis converts input data into set of statistically independent features. The proposed polynomial adaline has been applied to classify the patterns. The simulation results shows that the proposed polynomial adaline has better performances of the classification for test patterns, in comparison with those using the conventional polynomial adaline. Also, it is affected less by the scope of the smoothing factor.

• JSTS:Journal of Semiconductor Technology and Science
• /
• 제13권1호
• /
• pp.34-42
• /
• 2013

Human faces have been broadly studied in digital image and video processing fields. An appearance-based method, the adaptive boosting learning algorithm using integral image representations has been successfully employed for face detection, taking advantage of the feature extraction's low computational complexity. In this paper, we propose a face-detection postprocessing method that equalizes instantaneous facial regions in an efficient hardware architecture for use in real-time multimedia applications. The proposed system requires low hardware resources and exhibits robust performance in terms of the movements, zooming, and classification of faces. A series of experimental results obtained using video sequences collected under dynamic conditions are discussed.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2007년도 하계종합학술대회 논문집
• /
• pp.361-362
• /
• 2007

We propose a contrast-controlled feature detection approach for steel radiograph image. X-ray images are low contrast, dark and high noise image. So, It is not simple to detect defects directly in automated radiography inspection system. Contrast enhancement, histogram equalization and median filter are the most frequently used techniques to enhance the X-ray images. In this paper, the adaptive control method based on contrast limited histogram equalization is compared with several histogram techniques. Through comparative analysis, CLAHE(contrast controlled adaptive histogram equalization) can enhance detection of defects better.

• Journal of information and communication convergence engineering
• /
• 제20권3호
• /
• pp.153-159
• /
• 2022

Plain radiographic analysis is the initial imaging modality for suspected small bowel obstruction. Among the many features that affect the diagnosis of small bowel obstruction (SBO), the presence of gas-filled or fluid-filled small bowel loops is the most salient feature that can be automatized by computer vision algorithms. In this study, we compare three frequently applied pixel-clustering algorithms for extracting gas-filled areas without human intervention. In a comparison involving 40 suspected SBO cases, the Possibilistic C-Means and Fuzzy C-Means algorithms exhibited initialization-sensitivity problems and difficulties coping with low intensity contrast, achieving low 72.5% and 85% success rates in extraction. The Adaptive Resonance Theory 2 algorithm is the most suitable algorithm for gas-filled region detection, achieving a 100% success rate on 40 tested images, largely owing to its dynamic control of the number of clusters.