Imtiaz A. Taj

Velocity and pressure-based partitions of horizontal and vertical trajectories for on-line signature verification

In general, shape of an on-line signature is used as a single discriminating feature. Sometimes shape of signature is used alone for verification purposes and sometimes it is used in combination with some other dynamic features such as velocity, pressure and acceleration. The shape of an on-line signature is basically formed due to the wrist and fingers movements where the wrist movement is represented by the horizontal trajectory and the movement of the fingers is represented by vertical trajectory. As the on-line signature is formed due to the combination of two movements that are essentially independent of each other, it will be more effective to use them as two separate discriminating features. Based on this observation, we propose to use these trajectories in isolation by first decomposing the pressure and velocity profiles into two partitions and then extracting the underlying horizontal and vertical trajectories. So the overall process can be thought as the process which exploits the inter-feature dependencies by decomposing signature trajectories depending upon pressure and velocity information and performs verification on each partition separately. As a result, we are able to extract eight discriminating features and among them the most stable discriminating feature is used in verification process. Further Principal Component Analysis (PCA) has been proposed to make the signatures rotation invariant. Experimental results demonstrate superiority of our approach in on-line signature verification in comparison with other techniques.

Cross-term elimination in Wigner distribution based on 2D signal processing techniques

An efficient method based on 2D signal processing techniques and fractional Fourier transform is presented to suppress interference terms of Wigner distribution (WD). The proposed technique computes Gabor transform (GT) of a multi-component signal to obtain a blurred time-frequency (t-f) image. Signal components in GT image are segmented using connected component segmentation and are filtered out using precise application of fractional Fourier transform. A crisp t-f representation is then obtained by computing the sum of products of WD and GT of the isolated signal components. The efficacy of the proposed technique is demonstrated using examples of synthetic signals and real-life bat signals. Proposed scheme gives satisfactory performance even when cross-terms of WD overlap auto-terms and computational cost analysis shows that it is more efficient than recent interference suppression techniques of comparable performance. Moreover, the proposed technique does not require any prior info regarding the nature of signal.

Efficient 2-fold contextual filtering approach for fingerprint enhancement

Automated personal authentication has become increasingly important in modern information driven society and in this regard fingerprint-based personal identification is considered to be the most effective tool. In order to ensure reliable fingerprint identification and improve fingerprint ridge structure, a novel fingerprint enhancement approach is presented based on local adaptive contextual filtering. The proposed enhancement technique is 2-fold as it involves processing both in frequency and spatial domain. The fingerprint image is first filtered in frequency domain and then local directional filtering in spatial domain is applied to obtain enhanced fingerprint. In order to determine the performance efficiency of the proposed enhancement technique, a comparative analysis of error rates on standard fingerprint databases has been presented with major contextual enhancement schemes. The results show the efficacy of the proposed scheme as compared with other contextual filtering techniques.

Cardiac disorder diagnosis based on ECG segments analysis and classification

In this study we present a new approach to analyze and classify ECG signals for diagnosis of five cardiac conditions. Instead of following the conventional approach of beat-to-beat classification, we classify cardiac rhythms/segments of ECG based on statistical and morphological features extracted from them.

Optimized implementation of motion compensation for H.264 decoder

Predictive coding is widely used in video codec standards, especially for applications requiring low bit-rate coding like video broadcasting, video streaming and video conferencing. Motion compensation is used as a part of the predictive coding to provide a prediction for the macroblock. Considering that the motion compensation in H.264/AVC takes about one-fourth of the decoding time, in this paper we propose an optimized implementation of motion compensation algorithm. The proposed implementation uses the macroblock (MB) mode of decoded macroblocks to improve performance degradation in data manipulation and single-instruction, multiple data (SIMD) to simplify linear interpolation. Experimental results show that this implementation significantly improves timing performance of motion compensation and the speed of overall decoder without loss in video quality. The results of implementation on a general purpose DSP are compared with other recent optimization techniques showing the efficacy of proposed technique.

The role of facial asymmetry in recognizing age-separated face images

Facial asymmetry based approach is proposed to classify age-separated face images.Facial asymmetry is measured and evaluated across temporal variations.A 3 D matching-scores space is built using holistic, local and asymmetric features.SVM is used as classifier to separate genuine and imposter classes in score space.Results show better performance of proposed approach compared to existing methods. Recognition of age-separated face images is a challenging and open research problem. In this paper we propose a facial asymmetry based matching-score space (MSS) approach for recognition of age-separated face images. Motivated by its discriminatory information, we evaluate facial asymmetry across small and large temporal variations and use asymmetric facial features to recognize age-separated face images. We extract three different facial features including holistic feature descriptors using Principal Component Analysis (PCA), local feature descriptors using Local Binary Patterns (LBP), and Densely Sampled Asymmetric Features (DSAF) to represent face images. Then we develop MSS to discriminate genuine and imposter classes using support vector machine (SVM) as a classifier. Experimental results on three widely used face aging databases, the FERET, MORPH and FG-NET, show that proposed approach has superior performance compared to some existing state-of-the-art approaches. Display Omitted

A multifaceted independent performance analysis of facial subspace recognition algorithms.

Face recognition has emerged as the fastest growing biometric technology and has expanded a lot in the last few years. Many new algorithms and commercial systems have been proposed and developed. Most of them use Principal Component Analysis (PCA) as a base for their techniques. Different and even conflicting results have been reported by researchers comparing these algorithms. The purpose of this study is to have an independent comparative analysis considering both performance and computational complexity of six appearance based face recognition algorithms namely PCA, 2DPCA, A2DPCA, (2D)2PCA, LPP and 2DLPP under equal working conditions. This study was motivated due to the lack of unbiased comprehensive comparative analysis of some recent subspace methods with diverse distance metric combinations. For comparison with other studies, FERET, ORL and YALE databases have been used with evaluation criteria as of FERET evaluations which closely simulate real life scenarios. A comparison of results with previous studies is performed and anomalies are reported. An important contribution of this study is that it presents the suitable performance conditions for each of the algorithms under consideration.

Applying centroid based adjustment to kernel based object tracking for improving localization

In recent studies kernel based object tracking (KBOT) using Bhattacharya coefficient as similarity measure is shown to be robust and efficient object tracking technique. Image histogram provides a compact summarization of the distribution of data in an image. Due to computational efficiency; histogram has been successfully applied in KBOT based tracking algorithms. However without spatial or shape information, similar objects of different color may be indistinguishable based solely on histogram comparisons. The application of meanshift algorithm (the core of KBOT) on 1-D low level features of histogram may converge to false local maxima and cause inaccuracy of target localization. In this paper we presented a robust and efficient tracking approach using structural features along with histogram based Bhattacharya coefficient similarity measure for tracking non rigid objects. It is proposed that integrating the edge based target information as post processing step for updating estimated mean shift centroid in KBOT improves the localization problem. Experimental results show the updated algorithm has achieve more precise tracking results as compared to original kernel based object tracking

Creation and selection of most stable discriminating features for on-line signature verification

In this paper, we propose a method for exploiting the dependence of four dynamic features (horizontal trajectory, vertical trajectory, pressure and velocity) upon angles of each sample point from the mean of the signature. We have achieved this purpose by decomposing each of these four dynamic features into two partitions. As a result we obtain eight discriminating features, out of which we have selected two most stable features on the basis of training signatures scatter on two dimensional dissimilarity space. These two most stable features are then employed for verification purpose. Result shows the promise of our proposed technique.

Fingerprint frequency normalisation and enhancement using two-dimensional short-time Fourier transform analysis

A fingerprint image with non-uniform ridge frequencies can be considered as a two-dimensional dynamic signal. A non-uniform stress on the sensing area applied during fingerprint acquisition may result in a non-linear distortion that disturbs the local frequency of ridges adversely affecting the matching performance. This study presents a new approach based on Short time Fourier transform analysis and local adaptive contextual filtering for frequency distortion removal and enhancement. In the proposed approach, the fingerprint image is divided into sub-images to determine local dominant frequency and orientation. Gaussian Directional band pass filtering is then adaptively applied in frequency domain. These filtered sub-images are then combined in spatial domain using a novel technique to obtain the enhanced fingerprint image of high ridge quality and uniform inter-ridge distance. Simulation results show the efficacy of the proposed enhancement technique as compared to other well-known contextual filtering based enhancement techniques reported in the literature.