Hafiz Imtiaz

A template matching approach of one-shot-learning gesture recognition

This paper proposes a novel approach for gesture recognition from motion depth images based on template matching. Gestures can be represented with image templates, which in turn can be used to compare and match gestures. The proposed method uses a single example of an action as a query to find similar matches and thus termed one-shot-learning gesture recognition. It does not require prior knowledge about actions, foreground/background segmentation, or any motion estimation or tracking. The proposed method makes a novel approach to separate different gestures from a single video. Moreover, this method is based on the computation of space-time descriptors from the query video which measures the likeness of a gesture in a lexicon. These descriptor extraction methods include the standard deviation of the depth images of a gesture as well as the motion history image. Furthermore, two dimensional discrete Fourier transform is employed to reduce the effect of camera shift. The comparison is done based on correlation coefficient of the image templates and an intelligent classifier is proposed to ensure better recognition accuracy. Extensive experimentation is done on a very complicated and diversified dataset to establish the effectiveness of employing the proposed methods.

A wavelet-based dominant feature extraction algorithm for palm-print recognition

In this paper, a multi-resolution feature extraction algorithm for palm-print recognition is proposed based on two-dimensional discrete wavelet transform (2D-DWT), which efficiently exploits the local spatial variations in a palm-print image. The entire image is segmented into several small spatial modules and the effect of modularization in terms of the entropy content of the palm-print images has been investigated. A palm-print recognition scheme is developed based on extracting dominant wavelet features from each of these local modules. In the selection of the dominant features, a threshold criterion is proposed, which not only drastically reduces the feature dimension but also captures precisely the detail variations within the palm-print image. It is shown that, because of modularization of the palm-print image, the discriminating capabilities of the proposed features are enhanced, which results in a very high within-class compactness and between-class separability of the extracted features. The effect of using different mother wavelets for the purpose of feature extraction has been also investigated. A principal component analysis is performed to further reduce the feature dimension. From our extensive experimentations on different palm-print databases, it is found that the performance of the proposed method in terms of recognition accuracy and computational complexity is superior to that of some of the recent methods.

A face recognition scheme using wavelet-based local features

In this paper, a multi-resolution feature extraction algorithm for face recognition is proposed based on two-dimensional discrete wavelet transform (2D-DWT), which efficiently exploits the local spatial variations in a face image. For the purpose of feature extraction, instead of considering the entire face image, an entropy-based local band selection criterion is developed, which selects high-informative horizontal segments from the face image. In order to capture the local spatial variations within these high-informative horizontal bands precisely, dominant wavelet coefficients corresponding to each local region residing inside those horizontal bands are selected as features. In the selection of the dominant coefficients, a threshold criterion is proposed, which not only drastically reduces the feature dimension but also provides high within-class compactness and high between-class separability. Extensive experimentation is carried out upon standard face databases and a very high degree of recognition accuracy is achieved by the proposed method in comparison to those obtained by some of the existing methods.

A wavelet-domain local feature selection scheme for face recognition

A multi-resolution feature extraction algorithm for face recognition based on two-dimensional discrete wavelet transform (2D-DWT) is proposed in this paper, which exploits the local spatial variations in a face image effectively. Instead of considering the entire face image, an entropy-based local band selection criterion is developed, which selects high-informative horizontal bands from the face image for feature extraction. In order to capture the local spatial variations within these bands precisely, a histogram-based local dominant feature selection criterion is proposed. The proposed dominant wavelet coefficients, in terms of frequency of occurrence, corresponding to each local region residing inside those horizontal bands not only reduces the feature dimension drastically but also provides high within-class compactness and high between-class separability. Extensive experimentation is carried out upon standard face databases and in comparison to those obtained by some of the existing methods, a very high degree of recognition accuracy is achieved by the proposed method.

A DCT-based feature extraction algorithm for palm-print recognition

In this paper, a frequency domain feature extraction algorithm for palm-print recognition is proposed, which efficiently exploits the local spatial variations in a palm-print image. The entire image is segmented into several narrow-width spatial bands and a palm-print recognition scheme is developed based on extracting dominant spectral features from each of these bands using two-dimensional discrete cosine transform (2D-DCT). The proposed dominant spectral feature selection algorithm offers an advantage of very low feature dimension and it is capable of capturing precisely the detail variations within the palm-print image, which results in a very high within-class compactness and between-class separability of the extracted features. From our extensive experimentations on different palm-print databases, it is found that the performance of the proposed method in terms of recognition accuracy and computational complexity is superior to that of some of the recent methods.

A histogram-based dominant wavelet domain feature selection algorithm for palm-print recognition ☆

Abstract A feature extraction algorithm for palm-print recognition based on two dimensional discrete wavelet transform is proposed in this paper, which efficiently exploits the local spatial variations in a palm-print image. The palm-image is segmented into several spatial modules and a palm-print recognition scheme is developed, which extracts histogram-based dominant wavelet features from each of these local modules. The effect of modularization in terms of the entropy content of the palm-print images has been analyzed. The selection of dominant features for the purpose of recognition not only drastically reduces the feature dimension but also captures precisely the detail variations within the palm-print image. It is shown that, the modularization of the palm-print image enhances the discriminating capabilities of the proposed features and thereby results in high within-class compactness and between-class separability of the extracted features. Different types of Daubechies wavelets (in terms of use of number of vanishing moments, i.e., db1–db10) have been utilized for the purpose of feature extraction and the effect upon the recognition performance has been also investigated. In order to further reduce the feature dimension, a principal component analysis is performed. It is found from our extensive experimentations on different palm-print databases that the performance of the proposed method in terms of recognition accuracy and computational complexity is superior to that of some of the recent methods.

A Wavelet-Domain Local Dominant Feature Selection Scheme for Face Recognition

A multiresolution feature extraction algorithm for face recognition is proposed based on two-dimensional discrete wavelet transform (2D-DWT), which efficiently exploits the local spatial variations in a face image. For feature extraction, instead of considering the entire face image, an entropy-based local band selection criterion is developed, which selects high-informative horizontal segments from the face image. In order to capture the local spatial variations within these bands precisely, the horizontal band is segmented into several small spatial modules. The effect of modularization in terms of the entropy content of the face images has been investigated. Dominant wavelet coefficients corresponding to each module residing inside those bands are selected as features. A histogram-based threshold criterion is proposed to select dominant coefficients, which drastically reduces the feature dimension and provides high within-class compactness and high between-class separability. The effect of using different mother wavelets for the purpose of feature extraction has been also investigated. PCA is performed to further reduce the dimensionality of the feature space. Extensive experimentation is carried out upon standard face databases, and a very high degree of recognition accuracy is achieved by the proposed method in comparison to those obtained by some of the existing methods.

A face recognition scheme based on spectral domain cross-correlation function

This paper presents a simple yet efficient face recognition technique, where a feature extraction algorithm is proposed based on the principle of spectral domain cross-correlation. Instead of considering the spatial variation of a face image as a whole, first we concentrate on spectral variation of each row of the image individually, which is obtained using discrete cosine transform (DCT). As each of these rows could carry distinct characteristic of the face image, considering all of them would ensure the extraction of the variation in face geometry without discarding any information even in a minute scale. It is shown that the cross-correlations in the DCT-domain considering pairs of consecutive rows provide a signature of the particular face image reflecting the variation in the face geometry along the vertical direction. In a similar fashion, a horizontal signature can be obtained considering DCT-domain cross-correlations along consecutive columns. However, it is observed that in comparison to horizontal features, vertical features offer better within-class compactness and between-class separation. In the proposed method, these two signatures are utilized combinedly in order to obtain a distinguishable feature space. From extensive simulations on standard databases, it is found that the proposed feature extraction algorithm offers advantages of simple practical implementation with a high degree of face recognition accuracy.

Motion clustering-based action recognition technique using optical flow

A new technique for action clustering-based human action representation on the basis of optical flow analysis and random sample consensus (RANSAC) method is proposed in this paper. The apparent motion of the human subject with respect to the background is detected using optical flow analysis, while the RANSAC algorithm is used to filter out unwanted interested points. From the remaining key interest points, the human subject is localized and the rectangular area surrounding the human body is segmented both horizontally and vertically. Next, the percentage of change of interest points at every small blocks at the intersections of horizontal and vertical segments from frame to frame are accumulated in matrix form for different persons performing the same action. An average of all these matrices is used as a feature vector for that particular action. In addition, the change in the position of the person along X-axis and Y-axis are cumulated for an action and included in the feature vectors. For the purpose of recognition using the extracted feature vectors, a distance-based similarity measure and a support vector machine (SVM)-based classifiers have been exploited. From extensive experimentations upon benchmark motion databases, it is found that the proposed method offers not only a very high degree of accuracy but also computational savings.

An efficient face recognition algorithm based on frequency domain cross-correlation function

This paper presents a simple yet efficient feature extraction algorithm for face recognition based on the principle of spectral domain cross-correlation. Instead of considering the spatial data of a face image as a whole, spectral feature is extracted from the each row of the spatial data individually. As each of these rows bears distinct characteristic of the face image, considering row-wise Fourier domain representation of all of them ensures the extraction of detail variation in face geometry. It is shown that the cross-correlations obtained considering pairs of spectral representations of consecutive rows provide a signature of the particular face image reflecting the variation in the face geometry along the vertical direction. In a similar fashion, a horizontal signature can be obtained considering spectral cross-correlations along consecutive columns. In the proposed method, both of these horizontal and vertical signatures are utilized in order to obtain a distinguishable feature space for a particular person. It is found that the proposed feature extraction algorithm offers advantages of simple practical implementation with a high degree of face recognition accuracy.