Search

For a broad range of applications, hyperspectral image (HSI) classification is a hot topic in remote sensing, and convolutional neural network (CNN)-based methods are drawing increasing attention. However, to train millions of parameters in CNN requires a large number of labeled training samples, which are difficult to collect. A conventional Gabor filter can effectively extract spatial information with different scales and orientations without training, but it may be missing some important discriminative information. In this article, we propose the Gabor ensemble filter (GEF), a new convolutional filter to extract deep features for HSI with fewer trainable parameters. GEF filters each input channel by some fixed Gabor filters and learnable filters simultaneously, then reduces the dimensions by some learnable 1×1 filters to generate the output channels. The fixed Gabor filters can extract common features with different scales and orientations, while the learnable filters can learn some complementary features that Gabor filters cannot extract. Based on GEF, we design a network architecture for HSI classification, which extracts deep features and can learn from limited training samples. In order to simultaneously learn more discriminative features and an end-to-end system, we propose to introduce the local discriminant structure for cross-entropy loss by combining the triplet hard loss. Results of experiments on three HSI datasets show that the proposed method has significantly higher classification accuracy than other state-of-the-art methods. Moreover, the proposed method is speedy for both training and testing.

Domain adaptation (DA) and transfer learning with statistical property description is very important in image analysis and data classification. This article studies the domain adaptive feature representation problem for the heterogeneous data, of which both the feature dimensions and the sample distributions across domains are so different that their features cannot be matched directly. To transfer the discriminant information efficiently from the source domain to the target domain, and then enhance the classification performance for the target data, we first introduce two projection matrices specified for different domains to transform the heterogeneous features into a shared space. We then propose a joint kernel regression model to learn the regression variable, which is called feature translator in this article. The novelty focuses on the exploration of optimal experimental design (OED) to deal with the heterogeneous and nonlinear DA by seeking the covariance structured feature translators (CSFTs). An approximate and efficient method is proposed to compute the optimal data projections. Comprehensive experiments are conducted to validate the effectiveness and efficacy of the proposed model. The results show the state-of-the-art performance of our method in heterogeneous DA.

Conditional maximum mean discrepancy (CMMD) can capture the discrepancy between conditional distributions by drawing support from nonlinear kernel functions; thus, it has been successfully used for pattern classification. However, CMMD does not work well on complex distributions, especially when the kernel function fails to correctly characterize the difference between intraclass similarity and interclass similarity. In this paper, a new kernel learning method is proposed to improve the discrimination performance of CMMD. It can be operated with deep network features iteratively and thus denoted as KLN for abbreviation. The CMMD loss and an autoencoder (AE) are used to learn an injective function. By considering the compound kernel, that is, the injective function with a characteristic kernel, the effectiveness of CMMD for data category description is enhanced. KLN can simultaneously learn a more expressive kernel and label prediction distribution; thus, it can be used to improve the classification performance in both supervised and semisupervised learning scenarios. In particular, the kernel-based similarities are iteratively learned on the deep network features, and the algorithm can be implemented in an end-to-end manner. Extensive experiments are conducted on four benchmark datasets, including MNIST, SVHN, CIFAR-10, and CIFAR-100. The results indicate that KLN achieves the state-of-the-art classification performance.

Establishing correspondence between two given geometrical graph structures is an important problem in computer vision and pattern recognition. In this paper, we propose a robust graph matching (RGM) model to improve the effectiveness and robustness on the matching graphs with deformations, rotations, outliers, and noise. First, we embed the joint geometric transformation into the graph matching model, which performs unary matching over graph nodes and local structure matching over graph edges simultaneously. Then, the L 2,1 -norm is used as the similarity metric in the presented RGM to enhance the robustness. Finally, we derive an objective function which can be solved by an effective optimization algorithm, and theoretically prove the convergence of the proposed algorithm. Extensive experiments on various graph matching tasks, such as outliers, rotations, and deformations show that the proposed RGM model achieves competitive performance compared to the existing methods.

Image set recognition has been widely applied in many practical problems like real-time video retrieval and image caption tasks. Due to its superior performance, it has grown into a significant topic in recent years. However, images with complicated variations, e.g., postures and human ages, are difficult to address, as these variations are continuous and gradual with respect to image appearance. Consequently, the crucial point of image set recognition is to mine the intrinsic connection or structural information from the image batches with variations. In this work, a Discriminant Residual Analysis (DRA) method is proposed to improve the classification performance by discovering discriminant features in related and unrelated groups. Specifically, DRA attempts to obtain a powerful projection which casts the residual representations into a discriminant subspace. Such a projection subspace is expected to magnify the useful information of the input space as much as possible, then the relation between the training set and the test set described by the given metric or distance will be more precise in the discriminant subspace. We also propose a nonfeasance strategy by defining another approach to construct the unrelated groups, which help to reduce furthermore the cost of sampling errors. Two regularization approaches are used to deal with the probable small sample size problem. Extensive experiments are conducted on benchmark databases, and the results show superiority and efficiency of the new methods.

Local feature descriptor plays a key role in different image classification applications. Some of these methods such as local binary pattern and image gradient orientations have been proven effective to some extent. However, such traditional descriptors which only utilize single-type features, are deficient to capture the edges and orientations information and intrinsic structure information of images. In this paper, we propose a kernel embedding multiorientation local pattern (MOLP) to address this problem. For a given image, it is first transformed by gradient operators in local regions, which generate multiorientation gradient images containing edges and orientations information of different directions. Then the histogram feature which takes into account the sign component and magnitude component, is extracted to form the refined feature from each orientation gradient image. The refined feature captures more information of the intrinsic structure, and is effective for image representation and classification. Finally, the multiorientation refined features are automatically fused in the kernel embedding discriminant subspace learning model. The extensive experiments on various image classification tasks, such as face recognition, texture classification, object categorization, and palmprint recognition show that MOLP could achieve competitive performance with those state-of-the art methods.

Robust descriptor-based subspace learning with complex data is an active topic in pattern analysis and machine intelligence. A few researches concentrate the optimal design on feature representation and metric learning. However, traditionally used features of single-type, e.g., image gradient orientations (IGOs), are deficient to characterize the complete variations in robust and discriminant subspace learning. Meanwhile, discontinuity in edge alignment and feature match are not been carefully treated in the literature. In this paper, local order constrained IGOs are exploited to generate robust features. As the difference-based filters explicitly consider the local contrasts within neighboring pixel points, the proposed features enhance the local textures and the order-based coding ability, thus discover intrinsic structure of facial images further. The multimodal features are automatically fused in the most discriminant subspace. The utilization of adaptive interaction function suppresses outliers in each dimension for robust similarity measurement and discriminant analysis. The sparsity-driven regression model is modified to adapt the classification issue of the compact feature representation. Extensive experiments are conducted by using some benchmark face data sets, e.g., of controlled and uncontrolled environments, to evaluate our new algorithm.

In this paper, we propose a novel discriminative and compact coding (DCC) for robust face recognition. It introduces multiple error measurements into regression model. They collaborate to tune regression codes of different properties (sparsity, compactness, high discriminating ability, etc.), to further improve robustness and adaptivity of the regression model. We propose two types of coding models: 1) multiscale error measurements that produces sparse and highly discriminative codes and 2) inspires within-class collaborative representation that produces sparse and compact codes. The update of codes and the combination of different errors are automatically processed. DCC is also robust to the choice of parameters, producing stable regression residuals which are crucial to classification. Extensive experiments on benchmark datasets show that DCC has promising performance and outperforms other state-of-the-art regression models.

Lambertian model is a classical illumination model consisting of a surface albedo component and a light intensity component. Some previous researches assume that the light intensity component mainly lies in the large-scale features. They adopt holistic image decompositions to separate it out, but it is difficult to decide the separating point between large-scale and small-scale features. In this paper, we propose to take a logarithm transform, which can change the multiplication of surface albedo and light intensity into an additive model. Then, a difference (substraction) between two pixels in a neighborhood can eliminate most of the light intensity component. By dividing a neighborhood into subregions, edgemaps of multiple scales can be obtained. Then, each edgemap is multiplied by a weight that can be determined by an independent training scheme. Finally, all the weighted edgemaps are combined to form a robust holistic feature map. Extensive experiments on four benchmark data sets in controlled and uncontrolled lighting conditions show that the proposed method has promising results, especially in uncontrolled lighting conditions, even mixed with other complicated variations.

Face recognition under uncontrolled conditions, e.g., complex backgrounds and variable resolutions, is still challenging in image processing and computer vision. Although many methods have been proved well-performed in the controlled settings, they are usually of weak generality across different data sets. Meanwhile, several properties of the source domain, such as background and the size of subjects, play an important role in determining the final classification results. A transferrable representation learning model is proposed in this paper to enhance the recognition performance. To deeply exploit the discriminant information from the source domain and the target domain, the bioinspired face representation is modeled as structured and approximately stable characterization for the commonality between different domains. The method outputs a grouped boost of the features, and presents a reasonable manner for highlighting and sharing discriminant orientations and scales. Notice that the method can be viewed as a framework, since other feature generation operators and classification metrics can be embedded therein, and then, it can be applied to more general problems, such as low-resolution face recognition, object detection and categorization, and so forth. Experiments on the benchmark databases, including uncontrolled Face Recognition Grand Challenge v2.0 and Labeled Faces in the Wild show the efficacy of the proposed transfer learning algorithm.

In this paper, we propose a multilayer surface albedo (MLSA) model to tackle face recognition in bad lighting conditions, especially with reference images in bad lighting conditions. Some previous researches conclude that illumination variations mainly lie in the large-scale features of an image and extract small-scale features in the surface albedo (or surface texture). However, this surface albedo is not robust enough, which still contains some detrimental sharp features. To improve robustness of the surface albedo, MLSA further decomposes it as a linear sum of several detailed layers, to separate and represent features of different scales in a more specific way. Then, the layers are adjusted by separate weights, which are global parameters and selected for only once. A criterion function is developed to select these layer weights with an independent training set. Despite controlled illumination variations, MLSA is also effective to uncontrolled illumination variations, even mixed with other complicated variations (expression, pose, occlusion, and so on). Extensive experiments on four benchmark data sets show that MLSA has good receiver operating characteristic curve and statistical discriminating capability. The refined albedo improves recognition performance, especially with reference images in bad lighting conditions.

Face recognition with illumination or pose variation is a challenging problem in image processing and pattern recognition. A novel algorithm using band-reweighed Gabor kernel embedding to deal with the problem is proposed in this paper. For a given image, it is first transformed by a group of Gabor filters, which output Gabor features using different orientation and scale parameters. Fisher scoring function is used to measure the importance of features in each band, and then, the features with the largest scores are preserved for saving memory requirements. The reduced bands are combined by a vector, which is determined by a weighted kernel discriminant criterion and solved by a constrained quadratic programming method, and then, the weighted sum of these nonlinear bands is defined as the similarity between two images. Compared with existing concatenation-based Gabor feature representation and the uniformly weighted similarity calculation approaches, our method provides a new way to use Gabor features for face recognition and presents a reasonable interpretation for highlighting discriminant orientations and scales. The minimum Mahalanobis distance considering the spatial correlations within the data is exploited for feature matching, and the graphical lasso is used therein for directly estimating the sparse inverse covariance matrix. Experiments using benchmark databases show that our new algorithm improves the recognition results and obtains competitive performance.

Face recognition with occlusion is common in the real world. Inspired by the works of structured sparse representation, we try to explore the structure of the error incurred by occlusion from two aspects: the error morphology and the error distribution. Since human beings recognize the occlusion mainly according to its region shape or profile without knowing accurately what the occlusion is, we argue that the shape of the occlusion is also an important feature. We propose a morphological graph model to describe the morphological structure of the error. Due to the uncertainty of the occlusion, the distribution of the error incurred by occlusion is also uncertain. However, we observe that the unoccluded part and the occluded part of the error measured by the correntropy induced metric follow the exponential distribution, respectively. Incorporating the two aspects of the error structure, we propose the structured sparse error coding for face recognition with occlusion. Our extensive experiments demonstrate that the proposed method is more stable and has higher breakdown point in dealing with the occlusion problems in face recognition as compared to the related state-of-the-art methods, especially for the extreme situation, such as the high level occlusion and the low feature dimension.

Practical video scene and face recognition systems are sometimes confronted with low-resolution (LR) images. The faces may be very small even if the video is clear, thus it is difficult to directly measure the similarity between the faces and the high-resolution (HR) training samples. Traditional super-resolution (SR) methods based face recognition usually have limited performance because the target of SR may not be consistent with that of classification, and time-consuming SR algorithms are not suitable for real-time applications. In this paper, a new feature extraction method called Coupled Kernel Embedding (CKE) is proposed for LR face recognition without any SR preprocessing. In this method, the final kernel matrix is constructed by concatenating two individual kernel matrices in the diagonal direction, and the (semi-)positively definite properties are preserved for optimization. CKE addresses the problem of comparing multi-modal data that are difficult for conventional methods in practice due to the lack of an efficient similarity measure. Particularly, different kernel types (e.g., linear, Gaussian, polynomial) can be integrated into an uniformed optimization objective, which cannot be achieved by simple linear methods. CKE solves this problem by minimizing the dissimilarities captured by their kernel Gram matrices in the low- and high-resolution spaces. In the implementation, the nonlinear objective function is minimized by a generalized eigenvalue decomposition. Experiments on benchmark and real databases show that our CKE method indeed improves the recognition performance.

This paper studies a problem of image restoration that observed images are contaminated by Gaussian and impulse noise. Existing methods for this problem in the literature are based on minimizing an objective functional having the l(1) fidelity term and the Mumford-Shah regularizer. We present an algorithm on this problem by minimizing a new objective functional. The proposed functional has a content-dependent fidelity term which assimilates the strength of fidelity terms measured by the l(1) and l(2) norms. The regularizer in the functional is formed by the l(1) norm of tight framelet coefficients of the underlying image. The selected tight framelet filters are able to extract geometric features of images. We then propose an iterative framelet-based approximation/sparsity deblurring algorithm (IFASDA) for the proposed functional. Parameters in IFASDA are adaptively varying at each iteration and are determined automatically. In this sense, IFASDA is a parameter-free algorithm. This advantage makes the algorithm more attractive and practical. The effectiveness of IFASDA is experimentally illustrated on problems of image deblurring with Gaussian and impulse noise. Improvements in both PSNR and visual quality of IFASDA over a typical existing method are demonstrated. In addition, Fast_IFASDA, an accelerated algorithm of IFASDA, is also developed.