S. Noushath

(2D)2LDA: An efficient approach for face recognition

Although 2DLDA algorithm obtains higher recognition accuracy, a vital unresolved problem of 2DLDA is that it needs huge feature matrix for the task of face recognition. To overcome this problem, this paper presents an efficient approach for face image feature extraction, namely, (2D)^2LDA method. Experimental results on ORL and Yale database show that the proposed method obtains good recognition accuracy despite having less number of coefficients.

Letters: Diagonal Fisher linear discriminant analysis for efficient face recognition

In this paper, a novel subspace method called diagonal Fisher linear discriminant analysis (DiaFLD) is proposed for face recognition. Unlike conventional principal component analysis and FLD, DiaFLD directly seeks the optimal projection vectors from diagonal face images without image-to-vector transformation. The advantage of the DiaFLD method over the standard 2-dimensional FLD (2DFLD) method is, the former seeks optimal projection vectors by interlacing both row and column information of images while the latter seeks the optimal projection vectors by using only row information of images. Our test results show that the DiaFLD method is superior to standard 2DFLD method and some existing well-known methods.

Robust Unconstrained Handwritten Digit Recognition using Radon Transform

The performance of a character recognition system depends heavily on what features are being used. Though many kinds of features have been developed and their test performances on standard database have been reported, there is still room to improve the recognition rate by developing improved features. In this paper, we propose a novel system based on radon transform for handwritten digit recognition. We have used radon function which represents an image as a collection of projections along various directions. The resultant feature vector by applying this method is the input for the classification stage. A nearest neighbor classifier is used for the subsequent recognition purpose. A test performed on the MNIST handwritten numeral database and on Kannada handwritten numerals demonstrate the effectiveness and feasibility of the proposed method

Multilingual OCR system for South Indian scripts and English documents: An approach based on Fourier transform and principal component analysis

Character recognition lies at the core of the discipline of pattern recognition where the aim is to represent a sequence of characters taken from an alphabet [Kasturi, R., Gorman, L.O., Govindaraju, V., 2002. Document image analysis: a primer. Sadhana 27 (Part 1), 3-22]. Though many kinds of features have been developed and their test performances on standard database have been reported, there is still room to improve the recognition rate by developing improved features. In this paper, we present a multilingual character recognition system for printed South Indian scripts (Kannada, Telugu, Tamil and Malayalam) and English documents. South Indian languages are most popular languages in India and around the world. The proposed multilingual character recognition is based on Fourier transform and principal component analysis (PCA), which are two commonly used techniques of image processing and recognition. PCA and Fourier transforms are classical feature extraction and data representation techniques widely used in the area of pattern recognition and computer vision. Our experimental results show the good performance over the data sets considered.

SVD Based Algorithms for Robust Face and Object Recognition in Robot Vision Applications

Contemporary society is highly networked and thus biometrics based surveillance has paramount importance for various security based reasons. The automatic, remote and robot vision based system are being deployed in a large way [11, 12]. The success of these schemes is highly dependent on robust algorithms for both face and object recognition. In this paper, we propose a very robust approach to face/object recognition based on Singular Value Decomposition (SVD). We first provide technical reasons to substantiate the claims made by T.Yuan et al [1], then we provide appropriate reasons for the robust behavior of SVD and finally, we corroborate the proposed concepts through extensive experiments on two standard databases which includes face and objects under both clean and noise conditions.

Fisher Linear Discriminant Analysis and Connectionist Model for Efficient Image Recognition

Subspace analysis is an effective technique for dimensionality reduction, which aims at finding a low-dimensional space of high-dimensional data. Fisher linear discriminant analysis (FLD) and Neural Networks are commonly used techniques of image processing and recognition. In this paper, a new scheme of image feature extraction namely, the FLD and Generalized Regression Neural Networks (GRNN) approach for image recognition is presented. Linear discriminant analysis is usually performed to investigate differences among multivariate classes, to determine which attributes discriminate between the classes, and to determine the most parsimonious way to distinguish among classes. Two-dimensional linear discriminative features and GRNN are used to perform the classification. Experiments on the image database (Face, Object and Character) demonstrate the effectiveness and feasibility of the proposed method.

Mixture-of-Laplacian faces and its application to face recognition

The locality preserving projection (LPP), known as Laplacianfaces, was recently proposed as a transformation technique of mapping which optimally preserves the neighborhood structure of the dataset. In this paper, an efficient method for face recognition called mixture-of-Laplacianfaces (or LPP mixture model) is proposed, which obtains several sets of Laplacianfaces through Expectation-Maximization (EM) learning of Gaussian Mixture Models (GMM). Experiments carried out by using this on ORL, FERET and COIL-20 indicate superior performance as compared with method based on Laplacianfaces and other contemporary subspace methods.

Some Issues on Choices of Modalities for Multimodal Biometric Systems

Biometrics-based authentication has advantages over other mechanisms, but there are several variabilities and vulnerabilities that need to be addressed. No single modality or combinations of modalities can be applied universally that is best for all applications. This paper deliberates different combinations of physiological biometric modalities with different levels of fusion. In our experiments, we have selected Face, Palmprint, Finger Knuckle Print, Iris, and Handvein modalities. All the modalities are of image type and publicly available, comprising at least 100 users. Proper selection of modalities for fusion can yield desired level of performance. Through our experiments it is learnt that a multimodal system which is considered just by increasing number of modalities by fusion would not yield the desired level of performance. Many alternate options for increased performance are presented.

Unconstrained Handwritten Digit Recognition: Experimentation on MNIST Database

The performance of a character recognition system depends heavily on what features are being used. Though many kinds of features have been developed and their test performances on standard database have been reported, there is still room to improve the recognition rate by developing improved features. In this paper, we propose a Two-dimensional Principal Component Analysis (2D-PCA) for efficient handwritten digit recognition. 2D-PCA is based on 2D image matrices rather than 1D vectors so that image matrix does not need to be transform into a vector prior to feature extraction as done in PCA.1 For subsequent classification purpose we have used Generalized Regression Neural Network (GR.NN). A test performed on the MNIST handwritten numeral database showed a better recognition rate is among the best on the MNIST database.

Divide-and-Conquer Strategy Incorporated Fisher Linear Discriminant Analysis: An Efficient Approach for Face Recognition

Fisher linear discriminant analysis (FLD) is one of the most popular feature extraction methods in patter recognition and can obtain a set of so-called projection directions such that the ratio of the between-class and the within-class scatter matrices reaches its maximum. However in reality, the dimension of the patterns will be so high that the conventional way of obtaining Fisher projections makes the computational task a tedious one. To alleviate this problem, in this paper, divide-and-conquer strategy incorporated FLD (dcFLD) is presented with two objectives: one is to sufficiently utilize the contribution made by local parts of the whole image and the other is to still follow the same simple mathematical formulation as FLD. In contrast to the traditional FLD method, which operates directly on the whole pattern represented as a vector, dcFLD first divides the whole pattern into a set of subpatterns and acquires a set of projection vectors for each partition to extract corresponding local sub-features. These local sub-features are then conquered to obtain global features. Experimental results on several image databases comprising of face and object reveal the feasibility and effectiveness of the proposed method