• ETRI Journal
• /
• 제30권5호
• /
• pp.753-755
• /
• 2008

In this letter, we propose a new histogram equalization technique for feature compensation in speech recognition under noisy environments. The proposed approach combines a signal-to-noise-ratio-dependent feature reconstruction method and the class histogram equalization technique to effectively reduce the acoustic mismatch present in noisy speech features. Experimental results from the Aurora 2 task confirm the superiority of the proposed approach for acoustic feature compensation.

• 센서학회지
• /
• 제20권6호
• /
• pp.388-392
• /
• 2011

In this paper, a face detection method for interactive TV control system using a new feature, edge histogram feature, with a support vector machine(SVM) in the near-infrared(NIR) images is proposed. The edge histogram feature is extracted using 16-directional edge intensity and a histogram. Compared to the previous method using local binary pattern(LBP) feature, the proposed method using edge histogram feature has better performance in both smaller feature size and lower equal error rate(EER) for face detection experiments in NIR databases.

• 대한음성학회지:말소리
• /
• 제55권
• /
• pp.131-142
• /
• 2005

A mismatch between the training and the test conditions often causes a drastic decrease in the performance of the speech recognition systems. In this paper, non-linear transformation techniques based on histogram equalization in the acoustic feature space are studied for reducing the mismatched condition. The purpose of histogram equalization(HEQ) is to convert the probability distribution of test speech into the probability distribution of training speech. While conventional histogram equalization methods consider only the probability distribution of a test speech, for noise-corrupted test speech, its probability distribution is also distorted. The transformation function obtained by this distorted probability distribution maybe bring about miss-transformation of feature vectors, and this causes the performance of histogram equalization to decrease. Therefore, this paper proposes a new method of calculating noise-removed probability distribution by using assumption that the CDF of noisy speech feature vectors consists of component of speech feature vectors and component of noise feature vectors, and this compensated probability distribution is used in HEQ process. In the AURORA-2 framework, the proposed method reduced the error rate by over $44\%$ in clean training condition compared to the baseline system. For multi training condition, the proposed methods are also better than the baseline system.

• 대한전자공학회논문지SP
• /
• 제42권4호
• /
• pp.119-126
• /
• 2005

본 논문에서는 영상의 블록 분류 특성에 적응적인 대표 컬러 히스토그램 (representative color histogram)과 방향성 패턴 히스토그램 (directional pattern histogram)을 이용한 새로운 내용 기반 영상 검색 방법 (content-based image retrieval)을 제안한다. 제안한 방법에서는 영상을 일정한 크기의 블록으로 나누고, 분할된 블록의 분류 특성에 따라 컬러와 패턴 특징 벡터를 추출한다. 먼저 분할된 블록을 채도 (saturation)에 따라 휘도 블록 또는 컬러 블록으로 분류한 후, 휘도 블록에 대해서는 블록 평균휘도 쌍의 히스토그램을 구하고, 컬러 블록에 대해서는 블록 평균 컬러 쌍 히스토그램을 구함으로써 블록 분류 특징에 따라 컬러 특징 벡터를 추출한다. 또한 블록 휘도 변화의 기울기 (gradient)를 계산하여 방향성 분류를 행한 후 히스토그램을 계산함으로써 블록 방향성 패턴 특징을 추출한다. 본 논문에서 제안한 영상 검색 방법의 성능을 평가하기 위해서 컴퓨터 모의실험을 행한 결과 제안한 방법이 기존의 방법들보다 정확도 (precision) 및 특징 벡터 차원 (feature vector dimension) 크기 등의 객관적인 측면에서 우수함을 확인하였다.

• 정보통신설비학회논문지
• /
• 제9권2호
• /
• pp.73-79
• /
• 2010

The most popular technique for image retrieval in a heterogeneous collection of color images is the comparison of images based on their color histogram. The color histogram describes the distribution of colors in the color space of a color image. In the most image retrieval systems, the color histogram is used to compute similarities between the query image and all the images in a database. But, small changes in the resolution, scaling, and illumination may cause important modifications of the color histogram, and so two color images may be considered to be very different from each other even though they have completely related semantics. A new method of color feature representation based on the 3-dimensional RGB color map is proposed to improve the defects of the color histogram. The proposed method is based on the three 2-dimensional projection map evaluated by projecting the RGB color space on the RG, GB, and BR surfaces. The experimental results reveal that the proposed is less sensitive to small changes in the scene and that achieve higher retrieval performances than the traditional color histogram.

• Journal of information and communication convergence engineering
• /
• 제2권2호
• /
• pp.123-127
• /
• 2004

The most popular technique for image retrieval in a heterogeneous collection of color images is the comparison of images based on their color histogram. The color histogram describes the distribution of colors in the color space of a color image. In the most image retrieval systems, the color histogram is used to compute similarities between the query image and all the images in a database. But, small changes in the resolution, scaling, and illumination may cause important modifications of the color histogram, and so two color images may be considered to be very different from each other even though they have completely related semantics. A new method of color feature representation based on the 3-dimensional RGB color map is proposed to improve the defects of the color histogram. The proposed method is based on the three 2-dimensional projection map evaluated by projecting the RGB color space on the RG, GB, and BR surfaces. The experimental results reveal that the proposed is less sensitive to small changes in the scene and that achieve higher retrieval performances than the traditional color histogram.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제13권10호
• /
• pp.5197-5218
• /
• 2019

Large-scale retrieval algorithm is problem for visual analyses applications, along its research track. In this paper, we propose a high-efficiency region division-based image retrieve approaches, which fuse low-level local color histogram feature and texture feature. A novel image region division is proposed to roughly mimic the location distribution of image color and deal with the color histogram failing to describe spatial information. Furthermore, for optimizing our region division retrieval method, an image descriptor combining local color histogram and Gabor texture features with reduced feature dimensions are developed. Moreover, we propose an extended Canberra distance method for images similarity measure to increase the fault-tolerant ability of the whole large-scale image retrieval. Extensive experimental results on several benchmark image retrieval databases validate the superiority of the proposed approaches over many recently proposed color-histogram-based and texture-feature-based algorithms.

• 한국컴퓨터산업학회논문지
• /
• 제6권2호
• /
• pp.265-270
• /
• 2005

히스토그램은 컬러공간의 특징 때문에 조명에 매우 민감하며, 이동된 빛의 강도를 가지고 있을때 유사성을 떨어뜨릴 가능성이 커지기 때문에, 본 논문에서는 히스토그램의 영역을 몇 개의 영역으로, 나눠, 그 영역들을 계산하는 HAC(Histogram Area Calculation)라 불리는 새로운 검색 방법을 소개한다. 제안한 방식은 현재 히스토그램이 가지고 있는 특성에 기반하여 히스토그램의 영역을 계산하고, 유사성을 매칭시킴으로써 명암도 변화에 대해서, 기존의 다른 전통적인 히스토그램 방법이나, 병합된 히스토그램 방법보다 제안한 방식의 성능이 훨씬 뛰어나다는 것을 보여준다.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2000년도 하계종합학술대회 논문집(4)
• /
• pp.194-197
• /
• 2000

Wavelet transform used for content-based image retrieval has good performance in texture image. Image features for content-based image retrieval are color, texture, and shape. In this paper, we use color feature extracted from HSI color space known as most similar vision system to human vision system and texture feature extracted from wavelet histogram which has multiresolution property. Proposed method is compared with HSI color histogram method and wavelet histogram method. It is shown better performance.

• 대한전자공학회논문지
• /
• 제24권5호
• /
• pp.856-862
• /
• 1987

An local adaptive image segmentatin algorithm for local feature detection and effective clustering of unimodal histogram shape are proposed. Local adaptive difference image and its histogram are obtained from the input image. The parameters are derived from the histogram and used for the segmentation based on relaxatin process. The results showed effective region segmentation and good noise cleaning for the ocular fundus fluorescein angiogram which has low contrast and unimodal histogram.