Alireza Bosaghzadeh

Exponential Local Discriminant Embedding and Its Application to Face Recognition

Local discriminant embedding (LDE) has been recently proposed to overcome some limitations of the global linear discriminant analysis method. In the case of a small training data set, however, LDE cannot directly be applied to high-dimensional data. This case is the so-called small-sample-size (SSS) problem. The classical solution to this problem was applying dimensionality reduction on the raw data (e.g., using principal component analysis). In this paper, we introduce a novel discriminant technique called “exponential LDE” (ELDE). The proposed ELDE can be seen as an extension of LDE framework in two directions. First, the proposed framework overcomes the SSS problem without discarding the discriminant information that was contained in the null space of the locality preserving scatter matrices associated with LDE. Second, the proposed ELDE is equivalent to transforming original data into a new space by distance diffusion mapping (similar to kernel-based nonlinear mapping), and then, LDE is applied in such a new space. As a result of diffusion mapping, the margin between samples belonging to different classes is enlarged, which is helpful in improving classification accuracy. The experiments are conducted on five public face databases: Yale, Extended Yale, PF01, Pose, Illumination, and Expression (PIE), and Facial Recognition Technology (FERET). The results show that the performances of the proposed ELDE are better than those of LDE and many state-of-the-art discriminant analysis techniques.

Adaptive graph construction using data self-representativeness for pattern classification

Graph construction from data constitutes a pre-stage in many machine learning and computer vision tasks, like semi-supervised learning, manifold learning, and spectral clustering. The influence of graph construction procedures on learning tasks and their related applications has only received limited study despite its critical impact on accuracy. State-of-the-art graphs are built via sparse coding adopting ?1 regularization. Those graphs exhibit good performance in many computer vision applications. However, the locality and similarity among instances are not explicitly used in the coding scheme. Furthermore, due to the use of ?1 regularization, these construction approaches can be computationally expensive. In this paper, we investigate graph construction using the data self-representativeness property. By incorporating a variant of locality-constrained linear coding (LLC), we introduce and derive four variants for graph construction. These variants adopt a two phase LLC (TPLLC). Compared with the recent ?1 graphs, our proposed objective function, associated with three variants, has an analytical solution, and thus, is more efficient. A key element of the proposed methods is the second phase of coding that allows data closeness, or locality, to be naturally incorporated. It performs a coding over some selected relevant samples and reinforces the individual regularization terms by exploiting the coefficients estimated in the first phase. Comprehensive experimental results using several benchmark datasets show that it can achieve or outperform existing state-of-the-art results. Furthermore, it is shown to be more efficient than the robust ?1 graph construction schemes.

Efficient Graph Construction for Label Propagation Based Multi-observation Face Recognition

Human-machine interaction is a hot topic nowadays in the communities of multimedia and computer vision. In this context, face recognition algorithms (used as primary cue for a persons identity assessment) work well under controlled conditions but degrade significantly when tested in real-world environments. Recently, graph-based label propagation for multi-observation face recognition was proposed. However, the associated graphs were constructed in an ad-hoc manner (e.g., using the KNN graph) that cannot adapt optimally to the data. In this paper, we propose a novel approach for efficient and adaptive graph construction that can be used for multi-observation face recognition as well as for other recognition problems. Experimental results performed on Honda video face database, show a distinct advantage of the proposed method over the standard graph construction methods.

Efficient dynamic graph construction for inductive semi-supervised learning

Most of graph construction techniques assume a transductive setting in which the whole data collection is available at construction time. Addressing graph construction for inductive setting, in which data are coming sequentially, has received much less attention. For inductive settings, constructing the graph from scratch can be very time consuming. This paper introduces a generic framework that is able to make any graph construction method incremental. This framework yields an efficient and dynamic graph construction method that adds new samples (labeled or unlabeled) to a previously constructed graph. As a case study, we use the recently proposed Two Phase Weighted Regularized Least Square (TPWRLS) graph construction method. The paper has two main contributions. First, we use the TPWRLS coding scheme to represent new sample(s) with respect to an existing database. The representative coefficients are then used to update the graph affinity matrix. The proposed method not only appends the new samples to the graph but also updates the whole graph structure by discovering which nodes are affected by the introduction of new samples and by updating their edge weights. The second contribution of the article is the application of the proposed framework to the problem of graph-based label propagation using multiple observations for vision-based recognition tasks. Experiments on several image databases show that, without any significant loss in the accuracy of the final classification, the proposed dynamic graph construction is more efficient than the batch graph construction.

Advances in Face Image Analysis: Theory and Applications

Description: Advances in Face Image Analysis: Theory and applications describes several approaches to facial image analysis and recognition. Eleven chapters cover advances in computer vision and pattern recognition methods used to analyze facial data. The topics addressed in this book include automatic face detection, 3D face model fitting, robust face recognition, facial expression recognition, face image data embedding, modelless 3D face pose estimation and image-based age estimation.

Object Classification using Hybrid Holistic Descriptors: Application to Building Detection in Aerial Orthophotos

We present a framework for automatic and accurate multiple detection of objects of interest from images using hybrid image descriptors. The proposed framework combines a powerful segmentation algorithm with a hybrid descriptor. The hybrid descriptor is composed by color histograms and several Local Binary Patterns based descriptors. The proposed framework involves two main steps. The first one consists in segmenting the image into homogeneous regions. In the second step, in order to separate the objects of interest and the image background, the hybrid descriptor of each region is classiied using machine learning tools and a gallery of training descriptors. To show its performance, the method is applied to extract building roofs from orthophotos. We provide evaluation performances over 100 buildings. The proposed approach presents several advantages in terms of applicability, suitability and simplicity. We also show that the use of hybrid descriptors lead to an enhanced performance

Locality Constrained Encoding Graph Construction and Application to Outdoor Object Classification

In this paper, we develop a new efficient graph construction algorithm that is useful for many learning tasks. Unlike the main stream for graph construction, our proposed data self-representativeness approach simultaneously estimates the graph structure and its edge weights through sample coding. Compared with the recent l1 graph that is based on sparse coding, our proposed objective function has an analytical solution (based on self-representativeness of data) and thus is more efficient. This paper has two main contributions. Firstly, we introduce the Two Phase Weighted Regularized Least Square (TPWRLS) graph construction. Secondly, the obtained data graph is used, in a semi-supervised context, in order to categorize detected objects in driving/urban scenes using Local Binary Patterns as image descriptors. The experiments show that the proposed method can outperform competing methods.

Parameterless Local Discriminant Embedding

In this paper, we propose a parameterless Local Discriminant Embedding. Recently, local discriminant embedding (LDE) method was proposed in order to tackle some limitations of the global linear discriminant analysis (LDA) method. LDE splits the graph Laplacian into two components: within-class adjacency graph and between-class adjacency graph to better characterize the discriminant property of the data. However, it is very difficult to set in advance the within- and between-class graphs. Our proposed LDE variant has two important characteristics: (i) while all graph-based manifold learning techniques (supervised and unsupervised) are depending on several parameters that require manual tuning, ours is parameter-free, and (ii) it adaptively estimates the local neighborhood surrounding each sample based on the data similarity. The resulting revisited LDE approach has been applied to the problem of model-less coarse 3D head pose estimation (person independent 3D pose estimation). It was tested on two large databases: FacePix and Pointing’04. It was conveniently compared with other linear techniques. The experimental results confirm that our method outperforms, in general, the existing ones.

Reducing circular hough transform parameters using morphological operations

This article introduces a new technique to reduce the parameters of the Circular Hough Transform (CHT). CHT is a well-known technique to locate circles in an image. One of the main drawbacks of CHT is its three-dimensional parameter space (location and radius of the circle) which makes this algorithm not memory efficient. In this article, based on morphological operations, we reduce this parameter space to two parameters which greatly improves its speed and memory. In the first step, we determine the radius of the circles using morphological operations. In the second step, only the location of the circle centers should be found. This trick will reduce the need for a third parameter of the CHT, hence can greatly reduce the consumed memory. Moreover, by using morphologically processed images, the images that we feed to CHT mainly have circles with a specific radius while most of other objects are removed. This trick improves the speed of the algorithm since the number of false edges is greatly reduced. Experimental results on different images prove that the proposed method can detect circles with a two-dimensional accumulator space.

Recognizing multiple observations using adaptive graph based label propagation

Recently, we introduced a robust and adaptive method for constructing sparse graphs. This method was termed Two Phase Weighted Regularized Least Square (TPWRLS) [6]. In this framework, the graph structure and its affinity matrix are simultaneously computed through a two phase sample coding. The second phase of coding utilizes adaptive sample pruning and re-weighting. In the context of graph-based semi-supervised label propagation, the obtained graph can achieve or outperform state-of-the art graph construction methods. In this paper, we present a performance study of the proposed method by considering two main aspects that were not addressed before. First, the new graph is exploited in order to tackle the problem of recognizing multiple images corresponding to the same category-a non straightforward scenario for supervised recognition techniques. Second, a performance evaluation on different image descriptor types is carried out. Experiments are conducted on three public image datasets: two face datasets and one handwritten digit dataset. These experiments show that in addition to its superiority over competing graph construction methods, the proposed method can easily solve the label inference of multiple observations and can work with several types of image descriptors and scenes.