M. Raman Kumar
Other affiliations: Kakatiya Institute of Technology and Science
Bio: M. Raman Kumar is an academic researcher from VIT University. The author has contributed to research in topic(s): Facial recognition system & Wavelet. The author has an hindex of 4, co-authored 6 publication(s) receiving 24 citation(s). Previous affiliations of M. Raman Kumar include Kakatiya Institute of Technology and Science.
Topics: Facial recognition system, Wavelet, Feature extraction, Kernel principal component analysis, Wavelet transform
03 Nov 2011-
TL;DR: A preprocessing scheme using Singular Value Decomposition and Histogram Equalization to enhance and facilitate illumination invariant face recognition and is compared with PCA, KPCA without any preprocessing.
Abstract: Uncontrolled lighting Conditions poses obstacle to face recognition. To deal with this problem, this paper proposes a preprocessing scheme using Singular Value Decomposition and Histogram Equalization to enhance and facilitate illumination invariant face recognition. The proposed method first generates synthetic image using Histogram equalization. Original and synthetic images are singular value decomposed; from the estimates of singular values enhanced image is reconstructed. Enhanced image is discrete wavelet decomposed (Haar & Db4) in to different frequency sub bands (LL, LH, HL, HH). The LL sub band is the best approximation of original image with lower-dimensional space and is used as biometric template. Pose Invariant Feature vectors are extracted from this template using Kernel Principal Component Analysis (KPCA). To show the performance, the proposed method is tested on YaleB, ORL benchmarking Databases. The results obtained show the impact of the method and is compared with PCA, KPCA without any preprocessing.
01 Dec 2011-
TL;DR: A novel method of De-noising EOG signals using Dual Tree complex wavelet transform (DT-CWT) is proposed and is best suitable for real time EOG based applications like human-machine communication devices for disabled persons, eye movement analysis and gaming applications.
Abstract: Human activities are recognized from the Electrooculogram (EOG) signal generated from the movement of eye. Hence early, accurate preprocessing of EOG signals is important. In recent years, this became an active area of research. The EOG signal captured using acquisition device is corrupted with the noise and device intrinsic, thus pre processing (noise reduction) is first and foremost step in any further analysis & activity recognition. In this paper a novel method of De-noising EOG signals using Dual Tree complex wavelet transform (DT-CWT) is proposed. The Denoising results obtained are compared with conventional wavelet (DWT) de-noising method. To demonstrate the efficacy of the proposed method, SNR calculations and the statistical analysis are evaluated. The proposed method is best suitable for real time EOG based applications like human-machine communication devices for disabled persons, eye movement analysis and gaming applications.
21 Jul 2011-
TL;DR: Singular Value Decomposition is used to deal with surrounding illumination and wavelets are employed to aid the KPCA in capturing the Multi Scale Features there by making the System robust to pose and illumination variation.
Abstract: Biometric devices provide secure mechanism towards gaining access. One of the Biometric features is Face and the system implemented is Face Recognition system. The Classical Face Recognition System is implemented with Principal Component Analysis and is successful. PCA is a linear method of extracting the features in a lower dimension space and is severely affected by the Pose and surrounding illumination variation. To implement effective face recognition system, pose variation is to be considered and the problem is well addressed with Kernel PCA (nonlinear PCA). KPCA extracts features in a higher dimension space, there by the system is rugged to pose variation. The illumination variation is accounted for capture range of the front end device and its surrounding and is not dealt in KPCA. In this work Singular Value Decomposition is used to deal with surrounding illumination and wavelets are employed to aid the KPCA in capturing the Multi Scale Features there by making the System robust to pose and illumination variation. To show the performance, the proposed method is tested on YaleB, ORL Databases. The results obtained show the impact of the method and is compared with PCA, KPCA.
30 Sep 2019-
01 Jan 2020-
Abstract: The exceptional speed in increase of genomic data at public databases requires advanced computational tools to perform quick gene analysis. The tools can be devised with the aid of genomic signal processing. The pivotal task in genomic signal processing is numerical mapping. In numerical mapping, the string of nucleotides is transformed into discrete numerical sequence by assigning optimum mathematical descriptor to a nucleotide. The descriptor must be compatible with the further stages of genomic application in order to achieve high efficiency. In this work, a simple numerical mapping method is proposed in which the optimum descriptor value is obtained by applying Gray code concept. The proposed method is evaluated on benchmark databases HRM195 and ASP67 for an identification of protein coding region application. The proposed method exhibits improved exon prediction efficiency in terms of performance accuracy and equal error rate when compared with similar methods.
29 Apr 2014-IEEE Transactions on Nanobioscience
TL;DR: A hybrid de-noising method combining Discrete Wavelet Transformation (DWT) and an Adaptive Predictor Filter (APF) based on an adaptive autoregressive model for prediction of the waveform of signals in the ocular artifact zones that is well suited to applications in portable environments.
Abstract: Electroencephalogram (EEG) signals have a long history of use as a noninvasive approach to measure brain function. An essential component in EEG-based applications is the removal of Ocular Artifacts (OA) from the EEG signals. In this paper we propose a hybrid de-noising method combining Discrete Wavelet Transformation (DWT) and an Adaptive Predictor Filter (APF). A particularly novel feature of the proposed method is the use of the APF based on an adaptive autoregressive model for prediction of the waveform of signals in the ocular artifact zones. In our test, based on simulated data, the accuracy of noise removal in the proposed model was significantly increased when compared to existing methods including: Wavelet Packet Transform (WPT) and Independent Component Analysis (ICA), Discrete Wavelet Transform (DWT) and Adaptive Noise Cancellation (ANC). The results demonstrate that the proposed method achieved a lower mean square error and higher correlation between the original and corrected EEG. The proposed method has also been evaluated using data from calibration trials for the Online Predictive Tools for Intervention in Mental Illness (OPTIMI) project. The results of this evaluation indicate an improvement in performance in terms of the recovery of true EEG signals with EEG tracking and computational speed in the analysis. The proposed method is well suited to applications in portable environments where the constraints with respect to acceptable wearable sensor attachments usually dictate single channel devices.
01 Jan 2013-
TL;DR: Performance analysis of Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) for face recognition was carried out on various current PCA and LDA based face recognition algorithms using standard public databases.
Abstract: Analysing the face recognition rate of various current face recognition algorithms is absolutely critical in developing new robust algorithms In his paper we report performance analysis of Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) for face recognition This analysis was carried out on various current PCA and LDA based face recognition algorithms using standard public databases Among various PCA algorithms analyzed, Manual face localization used on ORL and SHEFFIELD database consisting of 100 components gives the best face recognition rate of 100%, the next best was 9970% face recognition rate using PCA based Immune Networks (PCA-IN) on ORL database Among various LDA algorithms analyzed, Illumination Adaptive Linear Discriminant Analysis (IALDA) gives the best face recognition rate of 989% on CMU PIE database, the next best was 98125% using Fuzzy Fisherface through genetic algorithm on ORL database In this paper we report performance analysis of various current PCA and LDA based algorithms for face recognition The evaluation parameter for the study is face recognition rate on various standard public databases The remaining of the paper is organized as follows: Section II provides a brief overview of PCA, Section III presents PCA algorithms analysed, Section IV provides brief overview of LDA, Section V presents LDA algorithms analysed Section VI presents performance analysis of various PCA and LDA based algorithms finally Section VII draws the conclusion II PRINCIPAL COMPONENT ANALYSIS (PCA)
TL;DR: The experimental results confirm the superiority of using S sub-band of SVD in terms of performance of the local descriptors over NIR face databases.
Abstract: Display Omitted Singular Value Decomposition (SVD) is used locally to enhance the local features.The local descriptors are computed over the S sub-band instead of raw values.The SVD and local descriptors are combined to characterize the NIR Face Images.The superiority is confirmed through NIR Face retrieval experiments.The results are presented over PolyU-NIR and CASIA-NIR face databases. From last decade, local descriptor such as Local Binary Pattern (LBP) is accepted as a very prominent feature descriptor for characterizing the images such as faces. The performance of such descriptors depends upon the local relationship of the image. The local relationship of the image can be utilized in more discriminative and robust way after some preprocessing as compared to the original image. The preprocessed images in the form of 4 sub-bands (i.e. S, U, V, and D sub-bands) are obtained by applying the Singular Value Decomposition (SVD) over the original image. The local descriptors are computed over these sub-bands (mainly S sub-band) and termed as the SVD based local descriptors. The performance of four local descriptors over SVD sub-bands are tested for near-infrared face retrieval using PolyU-NIR and CASIA-NIR face databases, and compared with the results obtained using descriptors without SVD sub-band. The experimental results confirm the superiority of using S sub-band of SVD in terms of performance of the local descriptors over NIR face databases.
01 Apr 2020-Biomedical Signal Processing and Control
TL;DR: A new numerical mapping method based on Walsh codes is proposed to detect the coding regions in eukaryotes and is efficient as it attains 94 % accuracy, 85 % sensitivity and 96 % specificity when tested on the benchmark C. Elegans gene sequence.
Abstract: The protein coding regions play a significant role for gene applications in genomic signal processing. Unlike prokaryotes, the coding regions in eukaryotes are arranged in a random manner. Owing to unequal lengths and low volume density of coding regions, the identification of coding regions makes cumbersome. In this work, a new numerical mapping method based on Walsh codes is proposed to detect the coding regions in eukaryotes. The Walsh code for each nucleotide is obtained using the statistical features of a DNA sequence. The proposed method uses static type of mapping to convert a string of DNA nucleotides into a numerical sequence. The numerical sequence is given as an input to the digital signal processing based spectrum identification tool to detect the existence of quasi-periodic components within the coding region. The advantage of our method is that it is simple to design and easy to represent. The performance of the proposed method has been tested on four benchmark databases and a random set of sequences collected from the National Center for Biological Information (NCBI) database. Furthermore, it has been compared with other state-of-the-art spectrum based numerical mapping methods for statistical features such as sensitivity, specificity and accuracy. The proposed method is efficient as it attains 94 % accuracy, 85 % sensitivity and 96 % specificity when tested on the benchmark C. Elegans gene sequence.
27 May 2019-Biomedizinische Technik
TL;DR: A robust EOG-based saccade recognition using the multi-channel convolutional independent component analysis (ICA) method and a constraint direction of arrival (DOA) algorithm that can automatically adjust the order of eye movement sources according to the constraint angle are proposed.
Abstract: Human activity recognition (HAR) is a research hotspot in the field of artificial intelligence and pattern recognition The electrooculography (EOG)-based HAR system has attracted much attention due to its good realizability and great application potential Focusing on the signal processing method of the EOG-HAR system, we propose a robust EOG-based saccade recognition using the multi-channel convolutional independent component analysis (ICA) method To establish frequency-domain observation vectors, short-time Fourier transform (STFT) is used to process time-domain EOG signals by applying the sliding window technique Subsequently, we apply the joint approximative diagonalization of eigenmatrix (JADE) algorithm to separate the mixed signals and choose the "clean" saccadic source to extract features To address the problem of permutation ambiguity in a case with a six-channel condition, we developed a constraint direction of arrival (DOA) algorithm that can automatically adjust the order of eye movement sources according to the constraint angle Recognition experiments of four different saccadic EOG signals (ie up, down, left and right) were conducted in a laboratory environment The average recognition ratios over 13 subjects were 9566% and 9733% under the between-subjects test and the within-subjects test, respectively Compared with "bandpass filtering", "wavelet denoising", "extended infomax algorithm", "frequency-domain JADE algorithm" and "time-domain JADE algorithm, the recognition ratios obtained relative increments of 46%, 349%, 285%, 281% and 291% (within-subjects test) and 491%, 343%, 221%, 224% and 228% (between-subjects test), respectively The experimental results revealed that the proposed algorithm presents robust classification performance in saccadic EOG signal recognition
Related Authors (1)
Author's H-index: 4