Recently, some spectral–spatial band selection (BS) strategies have become hugely popular as they fuse the spectral information of the pixels and the spatial relationship with the neighboring pixels to enhance the performance of classification methods. However, being unsupervised in nature, these methods do not utilize the class information of the training samples which could substantially empower the capabilities of such spectral–spatial BS methods. To circumvent this limitation, a supervised spectral–spatial BS method based on component loadings obtained from the principal components of spectral–spatial principal component analysis (PCA) and using a novel super-pixel based graph Laplacian embedding is proposed. The methodology attempts to unify the two strategies of dimensionality reduction, i.e., BS and feature extraction (FE), so that the benefits from both of them can be combined. The importance of each band is estimated in terms of its component loadings along the principal components which are estimated from a unified objective function consisting of three terms: data fidelity, classification error term, and spatial prior. Additionally, the spatial relationship among the neighboring samples is characterized using a novel superpixel-based graph model. An objective comparison of the proposed approach with several widely used, state-of-the-art BS methods demonstrates a significant improvement in the classification accuracy.

Although dimensionality reduction methods like PCA address the challenges associated with high dimensional hyperspec-tral data, they are less preferred as the extracted features are not easy to interpret. Moreover, bands with high variance mainly dominate the behavior of the top principal components. The proposed methodology exploits the potential of PCA to address these two challenges. Firstly, PCA is applied on spectrally segmented groups of highly correlated bands to ensure that it is not completely influenced by the high variance bands. Secondly, instead of using the principal components directly as features, the proposed approach selects the representative bands of each cluster by studying the correlation (component loadings) between bands and principal components. Experimental results indicate that the framework can proficiently select a subset of bands which combine both global as well subtle yet discriminative information for effective hyperspectral classification.

The poor spatial resolution of available ground data provides an opportunity to estimate solar irradiation from computational models based on the satellite data. The developed model employs various spectral bands of the satellite data and focuses on distinguish weather signatures, such as: dust, aerosols, fog, and clouds. These parameters affect the solar radiation availability all over the world. However, for a tropical country like India, which is potentially rich in solar energy resources, the study of these parameters is of crucial importance. A modified version of the HELIOSAT model is developed and an all-day index based on various bands of satellite data is introduced. Results are validated against the ground data recorded at various stations of the Indian Meterological Department as well as with the Meteosat East data.

Extracting inherent patterns from large data using decompositions of data matrix by a sampled subset of exemplars has found many applications in machine learning. We propose a computationally efficient algorithm for adaptive exemplar sampling, called fast exemplar selection (FES). The proposed algorithm can be seen as an efficient variant of the oASIS algorithm [1]. FES iteratively selects incoherent exemplars based on the exemplars that are already sampled. This is done by ensuring that the selected exemplars forms a positive definite Gram matrix which is checked by exploiting its Cholesky factorization in an incremental manner. FES is a deterministic rank revealing algorithm delivering a tighter matrix approximation bound. Further, FES can also be used to exactly represent low rank matrices and signals sampled from a unions of independent subspaces. Experimental results show that FES performs comparable to existing methods for tasks such as matrix approximation, feature selection, outlier detection, and clustering.

Features derived using sparse representation (SR)-based approaches have been shown to yield promising results for speech recognition tasks. In most of the approaches, the SR corresponding to speech signal is estimated using a dictionary, which could be either exemplar based or learned. However, a single-level decomposition may not be suitable for the speech signal, as it contains complex hierarchical information about various hidden attributes. In this paper, we propose to use a multilevel decomposition (having multiple layers), also known as the deep sparse representation (DSR), to derive a feature representation for speech recognition. Instead of having a series of sparse layers, the proposed framework employs a dense layer between two sparse layers, which helps in efficient implementation. Our studies reveal that the representations obtained at different sparse layers of the proposed DSR model have complimentary information. Thus, the final feature representation is derived after concatenating the representations obtained at the sparse layers. This results in a more discriminative representation, and improves the speech recognition performance. Since the concatenation results in a high-dimensional feature, principal component analysis is used to reduce the dimension of the obtained feature. Experimental studies demonstrate that the proposed feature outperforms existing features for various speech recognition tasks.

Depth map sensed by low-cost active sensor is often limited in resolution, whereas depth information achieved from structure from motion or sparse depth scanning techniques may result in a sparse point cloud. Achieving a high-resolution (HR) depth map from a low resolution (LR) depth map or densely reconstructing a sparse non-uniformly sampled depth map are fundamentally similar problems with different types of upsampling requirements. The first problem involves upsampling in a uniform grid, whereas the second type of problem requires an upsampling in a non-uniform grid. In this paper, we propose a new approach to address such issues in a unified framework, based on sparse representation. Unlike, most of the approaches of depth map restoration, our approach does not require an HR intensity image. Based on example depth maps, sub-dictionaries of exemplars are constructed, and are used to restore HR/dense depth map. In the case of uniform upsampling of LR depth map, an edge preserving constraint is used for preserving the discontinuity present in the depth map, and a pyramidal reconstruction strategy is applied in order to deal with higher upsampling factors. For upsampling of non-uniformly sampled sparse depth map, we compute the missing information in local patches from that from similar exemplars. Furthermore, we also suggest an alternative method of reconstructing dense depth map from very sparse non-uniformly sampled depth data by sequential cascading of uniform and non-uniform upsampling techniques. We provide a variety of qualitative and quantitative results to demonstrate the efficacy of our approach for depth map restoration.

We propose a novel computationally efficient hierarchical dictionary learning (HDL) approach for data-driven unmixing and functional connectivity analysis of functional magnetic resonance imaging (fMRI) data. It is shown that by simultaneously exploiting the sparsity of the spatial brain maps and the incoherence among their evolution in time or task functions, one can achieve better performance while overcoming the drawbacks of existing approaches. The task functions constituting the dictionary, are learned using a hierarchical subset selection approach. Here, to enforce incoherence among atoms, any new atom is selected from suitable training candidates if it does not lie in the column span of past selected atoms. Also, since the sparsity of spatial maps is generally unknown and affected due to acquisition artifacts, HDL doesn't make use of an implicit sparse coding stage while dictionary update. This makes HDL a very fast and efficient data-driven approach for fMRI analysis. Experimental results on synthetic and real fMRI datasets provide compelling evidences that HDL performs better than existing state-of-the-art methods.

We propose a novel classification framework to classify immunofluorescence images of HEp-2 cell specimens. We emphasize on using biologically motivated visual characteristics of classes, which we term as class-specific features. Given that the task involves less number of classes, a hierarchical verification based framework is employed, and is demonstrated to perform well. The current study focuses towards the classification of Homogeneous (H), Speckled (S) and Centromere (C) classes. The framework yields high classification rate with simple and efficient feature definitions. We also show encouraging performance for intermediate quality images, which represent early stage of diseases.

Abstract In this work, we propose sparse representation based features for speech units classification tasks. In order to effectively capture the variations in a speech unit, the proposed method employs multiple class specific dictionaries. Here, the training data belonging to each class is clustered into multiple clusters, and a principal component analysis (PCA) based dictionary is learnt for each cluster. It has been observed that coefficients corresponding to middle principal components can effectively discriminate among different speech units. Exploiting this observation, we propose to use a transformation function known as weighted decomposition (WD) of principal components, which is used to emphasize the discriminative information present in the PCA-based dictionary. In this paper, both raw speech samples and mel frequency cepstral coefficients (MFCC) are used as an initial representation for feature extraction. For comparison, various popular dictionary learning techniques such as K-singular value decomposition (KSVD), simultaneous codeword optimization (SimCO) and greedy adaptive dictionary (GAD) are also employed in the proposed framework. The effectiveness of the proposed features is demonstrated using continuous density hidden Markov model (CDHMM) based classifiers for (i) classification of isolated utterances of E-set of English alphabet, (ii) classification of consonant-vowel (CV) segments in Hindi language and (iii) classification of phoneme from TIMIT phonetic corpus.

We propose an example-based approach for enhancing resolution of range-images. Unlike most existing methods on range-image superresolution (SR), we do not employ a colour image counterpart for the range-image. Moreover, we use only a small set of range-images to construct a dictionary of exemplars. Considering the importance of edges in range-image SR, our formulation involves an edge-based constraint to better weight appropriate patches from the dictionary in a sparse-representation framework. Moreover, realizing the need for large up-sampling factors in case of range-images, we follow a hierarchical strategy for estimating the high-resolution range-images. We demonstrate that our strategy yields considerable improvements over the state-of-the-art approaches for range-image SR.

Quality of an image is associated with edge of the image. It is important to preserve the edge of the image while deriving high resolution (HR) image from low resolution (LR) image, also known as superresolution (SR) problem. This paper proposes an edge preserving constraint, which preserve the edge information of image by minimizing the differences between edges of LR image and the edges of the reconstructed image (down-sampled version), in sparse coding based SR problem. Partial edge evidences, derived using 1-D processing of image, are used separately in the constraints. The experimental results show that proposed approach preserves the edges of image as well as outperforms objectively the existing SR approaches.

Human faces are similar in structure with minor differences from person to person. These minor differences may average out while trying to synthesize the face image of a given person, or while building a model of face image in automatic face recognition. In this paper, we propose a template-matching approach for face verification, which neither synthesizes the face image nor builds a model of the face image. Template matching is performed using an edginess-based representation of the face image. The edginess-based representation of face images is computed using 1-D processing of images. An approach is proposed based on autoassociative neural network models to verify the identity of a person. The issues of pose and illumination in face verification are addressed.