• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제13권12호
• /
• pp.6028-6042
• /
• 2019

Since Sparse Nonnegative Matrix Factorization (SNMF) method can control the sparsity of the decomposed matrix, and then it can be adopted to control the sparsity of facial feature extraction and recognition. In order to improve the accuracy of SNMF method for facial feature recognition, new additive iterative rules based on the improved iterative step sizes are proposed to improve the SNMF method, and then the traditional multiplicative iterative rules of SNMF are transformed to additive iterative rules. Meanwhile, to further increase the sparsity of the basis matrix decomposed by the improved SNMF method, a threshold-sparse constraint is adopted to make the basis matrix to a zero-one matrix, which can further improve the accuracy of facial feature recognition. The improved SNMF method based on the additive iterative rules and threshold-sparse constraint is abbreviated as ASNMF, which is adopted to recognize the ORL and CK+ facial datasets, and achieved recognition rate of 96% and 100%, respectively. Meanwhile, from the results of the contrast experiments, it can be found that the recognition rate achieved by the ASNMF method is obviously higher than the basic NMF, traditional SNMF, convex nonnegative matrix factorization (CNMF) and Deep NMF.

• ETRI Journal
• /
• 제33권1호
• /
• pp.99-109
• /
• 2011

This paper concerns a robust real-time voice activity detection (VAD) approach which is easy to understand and implement. The proposed approach employs several short-term speech/nonspeech discriminating features in a voting paradigm to achieve a reliable performance in different environments. This paper mainly focuses on the performance improvement of a recently proposed approach which uses spectral peak valley difference (SPVD) as a feature for silence detection. The main issue of this paper is to apply a set of features with SPVD to improve the VAD robustness. The proposed approach uses a weighted voting scheme in order to take the discriminative power of the employed feature set into account. The experiments show that the proposed approach is more robust than the baseline approach from different points of view, including channel distortion and threshold selection. The proposed approach is also compared with some other VAD techniques for better confirmation of its achievements. Using the proposed weighted voting approach, the average VAD performance is increased to 89.29% for 5 different noise types and 8 SNR levels. The resulting performance is 13.79% higher than the approach based only on SPVD and even 2.25% higher than the not-weighted voting scheme.

• 감성과학
• /
• 제24권2호
• /
• pp.57-64
• /
• 2021

It has long been speculated that the visual system should use a coding strategy that takes advantage of statistical redundancies in images. But how such a coding strategy should manifest in neural responses has been less clear. Low-level image structure related to the power spectrum of natural images appears to be captured by a hard-wired efficient code in the retina of the fly and precortical structures like the LGN of cats that maximizes information content through the limited capacity channel of the optic nerve. But visual images are typically filled with higher-order structure beyond that captured by the power spectrum and visual cortex is not constrained by the same capacity limits as the optic nerve. Whether and how visual cortex can flexibly code for higher order redundancies is unknown. Here we show using psychophysical techniques that the neural response in early human visual cortex may be modulated by orientation redundancies in images such that a visual feature that is contained within a predictive pattern results in slower reaction times than a feature that deviates from a pattern, suggesting lower neural responses to predictable stimuli in the visual cortex. Our results point to a neural response in early visual cortex that is sensitive to global patterns and redundancies in visual images and is in marked contrast to standard models of cortical visual processing.

• 한국전자통신학회논문지
• /
• 제15권3호
• /
• pp.521-526
• /
• 2020

본 논문은 홍채영상의 이동불변의 특징값 을추출에 탁월한 고차 국소 자동 상관함수를 적용하여 25개의 특징 값을 입력 값으로 적용한 일반적인 HOLP 신경망에 특징 값 25개의 평균값을 추가한 개선된 HOLP 신경망을 구현하여 인식률을 확인하여 보았다. 종류가 상이한 딥러닝 구조들과 비교하였을 때 음성과 영상분야에서 탁월한 성능을 보이는 Back-Propagation 신경망과 특징 추출기와 분류기를 통합한 합성 곱 신경망을 활용하여 홍채인식의 인식률을 비교하여 보았다.

• 천문학회보
• /
• 제46권2호
• /
• pp.72.3-73
• /
• 2021

Baryon Acoustic Oscillations (BAO) are caused by acoustic density waves in the early universe and act as a standard ruler in the clustering pattern of galaxies in the late Universe. Measuring the BAO feature in the 2-point correlation function of a sample of galaxies allows us to estimate cosmological distances to the galaxies mean redshift, , which is important for testing and constraining the cosmology model. The BAO feature is also expected to appear in the higher order statistics. In this work we measure the generalized spatial N-point point correlation functions up to 4th order. We made measurements of the 2, 3, and 4-point correlation functions in the SDSS-III DR12 CMASS data, comprising of 777,202 galaxies. The errors and covariances matrices were estimated from 500 mock catalogues. We created a theoretical model for these statistics by measuring the N-point functions in halo catalogues produced by the approximate Lagrangian perturbation theory based simulation code, PINOCCHIO. We created simulations using initial conditions with and without the BAO feature. We find that the BAO is detected to high significance up to the 4-point correlation function.

• Structural Engineering and Mechanics
• /
• 제66권6호
• /
• pp.761-769
• /
• 2018

In this paper, we study the Carbon/Glass hybrid laminated composite plates, where the buckling behavior is examined using an accurate and simple refined higher order shear deformation theory. This theory takes account the shear effect, where shear deformation and shear stresses will be considered in determination of critical buckling load under different boundary conditions. The most interesting feature of this new kind of hybrid laminated composite plates is that the possibility of varying components percentages, which allows us for a variety of plates with different materials combinations in order to overcome the most difficult obstacles faced in traditional laminated composite plates like (cost and strength). Numerical results of the present study are compared with three-dimensional elasticity solutions and results of the first-order and the other higher-order theories issue from the literature. It can be concluded that the proposed theory is accurate and simple in solving the buckling behavior of hybrid laminated composite plates and allows to industrials the possibility to adjust the component of this new kind of plates in the most efficient way (reducing time and cost) according to their specific needs.

• Steel and Composite Structures
• /
• 제8권5호
• /
• pp.343-359
• /
• 2008

In the present study, an efficient method for the optimum design of three-dimensional (3D) steel framed structures is proposed. In this method, in addition to choosing the best position of columns based on architectural requirements, the optimum cross-sectional dimensions of elements are determined. The preliminary design variables are considered as the number of columns in structural plan, which are determined by a direct optimization method suitable for discrete variables, without requiring the evaluation of derivatives. After forming the geometry of structure, the main variables of the cross-sectional dimensions are evaluated, which satisfy the design constraints and also achieve the least-weight of the structure. To reduce the number of finite element analyses and the overall computational time, a new third order approximate function is introduced which employs only the diagonal elements of the higher order derivatives matrices. This function produces a high quality approximation and also, a robust optimization process. The main feature of the proposed techniques that the higher order derivatives are established by the first order exact derivatives. Several examples are solved and efficiency of the new approximation method and also, the proposed method for the best position of columns in 3D steel framed structures is discussed.

• 한국멀티미디어학회논문지
• /
• 제16권11호
• /
• pp.1281-1287
• /
• 2013

무분별한 복제 비디오를 막기 위하여 비디오 지문을 개발연구가 진행되고 있다. 이러한 비디오 지문들은 복제 비디오에서 발생되는 다양한 변화에 강인해야 하며 정확하게 구별될 수 있는 높은 분별력을 지녀야 한다. 일반적으로 비디오 지문들은 luminance(밝기), gradient(기울기), 그리고 DCT(주파수) 공간 등에서 주로 추출이 되고 있다. 그러나 아직 각 공간과 비디오 지문 사이에 실질적인 성능이 차이에 대한 연구가 부족하다. 따라서 본 논문에서는 각 공간에 따른 복제 비디오 검출 성능을 비교하기 위하여 강인함과 분별력에 기반한 복제 비디오 검출 실험을 진행하고 분석 하였다. 본 논문에서 동일한 패턴으로 각 공간에서 비디오 지문을 추출하고 각각 강인함과 분별력을 비교 한 후 최종적으로 복제 비디오 검출 실험을 진행하였다. 본 실험에서 DCT 공간에서 추출된 비디오 지문이 다른 공간보다 좀더 우수한 성능을 보여 주었는데 이는 해당 공간이 다른 비디오 지문들과 분별력이 가장 높았기 때문이다.

• 한국정보통신학회논문지
• /
• 제9권7호
• /
• pp.1513-1517
• /
• 2005

본 논문에서는 얼굴영역을 검출하기위해 얼굴 피부색을 보다 효과적으로 모델링하기 위한 피부색 특성을 고려하여 밝기 성분을 제거한 Red, Blue, Green 채널을 모두 사용하는 Hue, Cb, Cg의 M배i-Channel 피부색 모델을 사용한다. 얼굴영역을 분리한 영상에 Harr 웨이블릿을 이용한 에지영상 추출과 얼굴영역의 특징벡터를 구하기 위하여 26개의 특징벡터를 사용한 효율적인 고차 국소 자동 상관함수를 사용하였다. 계산된 특징벡터는 BP 신경망의 학습을 통하여 얼굴인식을 위한 데이터로 사용된다. 시뮬레이션을 통해 제안된 알고리즘에 의한 인식률향상과 속도 향상을 입증한다.

• 한국정보통신학회논문지
• /
• 제6권4호
• /
• pp.624-631
• /
• 2002

본 논문에서는 이진영상 데이터의 저작권 보호를 위한 디지털 워터마킹 방법을 제안하였다. 제안 방법은 먼저 이진영상을 기하학적 특징이 포함된 특징영역과 그 외의 일반영역으로 분리한다. 그리고 기하학적 특징이 포함된 이진영상의 특징영역을 보존하면서 인증을 위한 워터마크를 일반영역에만 삽입, 검출하는 효과 적인 워터마킹 방법을 연구하였다. 특징영역은 특수런을 사용한 런길이부호화를 이용하여 일반영역과 분리하였다. 워터마크의 비가시성을 위해 워터마크는 일반영역중에서 화소변화에 대한 민감도를 고려하여 삽입하였다. 제안 방법을 문자, 서명, 도장, 지문영상에 적용하여 성능을 평가하였다. 실험 결과 제안 방법은 원영상의 중요한 특징영역을 보존하고, 또한 워터마킹된 영상의 비가시성이 높음을 알 수 있었다.