• ETRI Journal
• /
• v.39 no.1
• /
• pp.21-29
• /
• 2017

This paper addresses the problem of unsupervised speech separation based on robust non-negative matrix factorization (RNMF) with ${\beta}$-divergence, when neither speech nor noise training data is available beforehand. We propose a robust version of non-negative matrix factorization, inspired by the recently developed sparse and low-rank decomposition, in which the data matrix is decomposed into the sum of a low-rank matrix and a sparse matrix. Efficient multiplicative update rules to minimize the ${\beta}$-divergence-based cost function are derived. A convolutional extension of the proposed algorithm is also proposed, which considers the time dependency of the non-negative noise bases. Experimental speech separation results show that the proposed convolutional RNMF successfully separates the repeating time-varying spectral structures from the magnitude spectrum of the mixture, and does so without any prior training.

• Journal of Ubiquitous Convergence Technology
• /
• v.2 no.1
• /
• pp.18-26
• /
• 2008

The invariance relation existing in the non-negative matrix factorization (NMF) is used for constructing robust image hashes in this work. The image is first re-scaled to a fixed size. Low-pass filtering is performed on the luminance component of the re-sized image to produce a normalized matrix. Entries in the normalized matrix are pseudo-randomly re-arranged under the control of a secret key to generate a secondary image. Non-negative matrix factorization is then performed on the secondary image. As the relation between most pairs of adjacent entries in the NMF's coefficient matrix is basically invariant to ordinary image processing, a coarse quantization scheme is devised to compress the extracted features contained in the coefficient matrix. The obtained binary elements are used to form the image hash after being scrambled based on another key. Similarity between hashes is measured by the Hamming distance. Experimental results show that the proposed scheme is robust against perceptually acceptable modifications to the image such as Gaussian filtering, moderate noise contamination, JPEG compression, re-scaling, and watermark embedding. Hashes of different images have very low collision probability. Tampering to local image areas can be detected by comparing the Hamming distance with a predetermined threshold, indicating the usefulness of the technique in digital forensics.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.52 no.5
• /
• pp.136-146
• /
• 2015

The active control of a vehicle based on vehicle recognition is one of key technologies for the intelligent vehicle, and the part-based image representation is necessary to recognize vehicles with only partial shapes of vehicles especially in urban scene where occlusions frequently occur. In this paper, we implemented a part-based image representation scheme using non-negative tensor factorization(NTF) and realized a robust vehicle recognition system using the NTF feature. The result shows that the proposed method gives more intuitive part-based representation and more robust recognition in urban scene.

• The Journal of the Acoustical Society of Korea
• /
• v.41 no.3
• /
• pp.326-334
• /
• 2022

In this paper, we propose speech enhancement algorithm as a pre-processing for robust speech recognition in noisy environments. Auxiliary-function-based Independent Vector Analysis (AuxIVA) is performed with weighted covariance matrix using time-varying variances with scaling factor from target masks representing time-frequency contributions of target speech. The mask estimates can be obtained using Neural Network (NN) pre-trained for speech extraction or diffuseness using Coherence-to-Diffuse power Ratio (CDR) to find the direct sounds component of a target speech. In addition, outputs for omni-directional noise are closely chained by sharing the time-varying variances similarly to independent subspace analysis or IVA. The speech extraction method based on AuxIVA is also performed in Independent Low-Rank Matrix Analysis (ILRMA) framework by extending the Non-negative Matrix Factorization (NMF) for noise outputs to Non-negative Tensor Factorization (NTF) to maintain the inter-channel dependency in noise output channels. Experimental results on the CHiME-4 datasets demonstrate the effectiveness of the presented algorithms.

• ETRI Journal
• /
• v.32 no.2
• /
• pp.345-347
• /
• 2010

In this letter, we present a new speech hash function based on the non-negative matrix factorization (NMF) of linear prediction coefficients (LPCs). First, linear prediction analysis is applied to the speech to obtain its LPCs, which represent the frequency shaping attributes of the vocal tract. Then, the NMF is performed on the LPCs to capture the speech's local feature, which is then used for hash vector generation. Experimental results demonstrate the effectiveness of the proposed hash function in terms of discrimination and robustness against various types of content preserving signal processing manipulations.

• ETRI Journal
• /
• v.36 no.1
• /
• pp.167-170
• /
• 2014

In this letter, we propose an unsupervised framework for speech noise reduction based on the recent development of low-rank and sparse matrix decomposition. The proposed framework directly separates the speech signal from noisy speech by decomposing the noisy speech spectrogram into three submatrices: the noise structure matrix, the clean speech structure matrix, and the residual noise matrix. Evaluations on the Noisex-92 dataset show that the proposed method achieves a signal-to-distortion ratio approximately 2.48 dB and 3.23 dB higher than that of the robust principal component analysis method and the non-negative matrix factorization method, respectively, when the input SNR is -5 dB.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.23 no.6
• /
• pp.682-689
• /
• 2019

In part-based image representation, the partial shapes of an object are represented as basis vectors, and an image is decomposed as a linear combination of basis vectors where the coefficients of those basis vectors represent the partial (or local) feature of an object. In this paper, a face recognition for occluded faces is proposed in which face images are represented using non-negative matrix factorization(NMF), one of part-based representation techniques, and recognized using an artificial neural network technique. Standard NMF, projected gradient NMF and orthogonal NMF were used in part-based representation of face images, and their performances were compared. Learning vector quantizer were used in the recognizer where Euclidean distance was used as the distance measure. Experimental results show that proposed recognition is more robust than the conventional face recognition for the occluded faces.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.41 no.6
• /
• pp.175-186
• /
• 2004

We present a hybrid approach of texture-based method and connected component (CC)-based method for text extraction in complex images. Two primary methods, which are mainly utilized in this area, are sequentially merged for compensating for their weak points. An automatically constructed MLP-based texture classifier can increase recall rates for complex images with small amount of user intervention and without explicit feature extraction. CC-based filtering based on the shape information using NMF enhances the precision rate without affecting overall performance. As a result, a combination of texture and CC-based methods leads to not only robust but also efficient text extraction. We also enhance the processing speed by adopting appropriate region marking methods for each input image category.

• 한국정보통신설비학회:학술대회논문집
• /
• 2009.08a
• /
• pp.143-146
• /
• 2009

There are some image hash methods, in the paper four image hash methods have been compared: FJLT (Fast Johnson- Lindenstrauss Transform), SVD (Singular Value Decomposition), NMF (Non-Negative Matrix Factorization), FP (Feature Point). From the compared result, FJLT method can't be used in the online. the search time is very slow because of the KNN algorithm. So FJLT method has been improved in the paper.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.37 no.7C
• /
• pp.554-564
• /
• 2012

The vehicle recognition consists of two steps; the vehicle region detection step and the vehicle identification step based on the feature extracted from the detected region. Features using linear transformations have the effect of dimension reduction as well as represent statistical characteristics, and show the robustness in translation and rotation of objects. Among the linear transformations, the NMF(Non-negative Matrix Factorization) is one of part-based representation. Therefore, we can extract NMF features with sparsity and improve the vehicle recognition rate by the representation of local features of a car as a basis vector. In this paper, we propose a feature extraction using NMF suitable for the vehicle recognition, and verify the recognition rate with it. Also, we compared the vehicle recognition rate for the occluded area using the SNMF(sparse NMF) which has basis vectors with constraint and LVQ2 neural network. We showed that the feature through the proposed NMF is robust in the urban scene where occlusions are frequently occur.