In spite of the overwhelming high-tech marvels and applications that rule our digital lives, the use of the handwritten signature is still recognized worldwide in government, personal and legal entities to be the most important behavioral biometric trait. A number of notable research approaches provide advanced results up to a certain point which allow us to assert with confidence that the performance attained by signature verification (SV) systems is comparable to those provided by any other biometric modality. Up to now, the mainstream trend for offline SV is shared between standard -or handcrafted- feature extraction methods and popular machine learning techniques, with typical examples ranging from sparse representation to Deep Learning. Recent progress in graph mining algorithms provide us with the prospect to re-evaluate the opportunity of utilizing graph representations by exploring corresponding graph features for offline SV. In this paper, inspired by the recent use of image visibility graphs for mapping images into networks, we introduce for the first time in offline SV literature their use as a parameter free, agnostic representation for exploring global as well as local information. Global properties of the sparsely located content of the shape of the signature image are encoded with topological information of the whole graph. In addition, local pixel patches are encoded by sequential visibility motifs-subgraphs of size four, to a low six dimensional motif profile vector. A number of pooling functions operate on the motif codes in a spatial pyramid context in order to create the final feature vector. The effectiveness of the proposed method is evaluated with the use of two popular datasets. The local visibility graph features are considered to be highly informative for SV; this is sustained by the corresponding results which are at least comparable with other classic state-of-the-art approaches.

The handwritten signature is perhaps the most accustomed way for the acknowledgement of the consent of an individual or the authentication of the identity of a person in numerous transactions. In addition, the authenticity of a questioned offline or static handwritten signature still poses a case of interest, especially in forensic related applications. A common approach in offline signature verification system is to apply several predetermined image analysis models. Consequently, any offline signature sample which originates from either authentic persons or forgers, utilizes a fixed feature extraction base. In this proposed study, the feature space and the corresponding projection values depend on the training samples only; thus the proposed method can be found useful in forensic cases. In order to do so, we reenter a groundbreaking unsupervised learning method named archetypal analysis, which is connected to effective data analysis approaches such as sparse coding. Due to the fact that until recently there was no efficient implementation publicly available, archetypal analysis had only few cases of use. However, a fast optimization scheme using an active set strategy is now available. The main goal of this work is to introduce archetypal analysis for offline signature verification. The output of the archetypal analysis of few reference samples is a set of archetypes which are used to form the base of the feature space. Then, given a set of archetypes and a signature sample under examination archetypal analysis and average pooling provides the corresponding features. The promising performance of the proposed approach is demonstrated with the use of an evaluation method which employs the popular CEDAR and MCYT75 signature datasets.

In this paper a new perspective is provided on the re-ranking problem, which is essential in pattern recognition and computer vision tasks. Items are efficiently organized using the minimal spanning tree (MST) and the orthogonal-MST graph and their similarity is calculated through an appropriate graph traversal method. The graph is augmented consecutively providing alternative paths, however not escaping the data manifold. The introduced method exploits the structure of the underlined manifold and is successfully applied (but not limited) to handwritten digits image database.

This work introduces a novel representation for the off-line handwritten signature by measuring its structural characteristics with a set of partially ordered grid features. Thirty-two binary symbols are delineated within the five-by-five pixel window and considered to be the alphabet of a probabilistic source. Subsequently and for practical purposes the whole set is appropriately organized into subsets of four symbols each. The new arrangement is used to detect the presence of simple or compound symbols in the signature image. The utilization of the partially ordered set or poset notion, intuitively arranges the binary feature extraction masks into first order chains, creating in this way a first order probabilistic description of the signature's structure that is characteristic of the motoric signature generating process. In order to estimate these probabilities, the first order searching strategy is limited to pixels neighbors having their grids centered to a predetermined Chebyshev distance of two. The utilized verification strategy relies on an SVM based classifier and the equal error rate figure. Our experimental scheme handles skilled forgery as well and derives verification results for two datasets, the GPDS300 and a proprietary one.

Indirect Immunofluorescence (IIF) followed by manual evaluation of the acquired slides from specialized personnel is the preferred laboratory technique used for the detection of Antinucleolar Antibodies (ANAs) in patient serum. In this procedure, several limitations appear and thus several automatic techniques have been proposed for the task of ANA detection. In this paper we propose a system for automatic classification of HEp-2 staining patterns, inspired by a recently proposed method for aggregating local image (SIFT) features into a compact and fixed length representation. More specifically we present a novel framework in which aggregated features are mapped into feature vectors in the dissimilarity space where the dimensionality of the descriptors is “naturally reduced”. The final descriptor is low dimensional, while evaluation on a recently published dataset yields state of the art results.

In this work we present a new approach to the extraction of features robust to focal mismatches, for the classification of biological particles characterized by 3 dimensional structures. We use SIFT descriptors in order to encode local gradient, fused with features derived from an introduced adaptive filterbank of Gabor filters. We have evaluated the proposed technique using a dataset consisting of 174 images of pollen grains from 29 species, acquired with a low-cost optical microscope in arbitrary focal planes. The proposed descriptor efficiently captures discriminative information by encoding the local inner and outer structure of the transparent pollens in a focus-tolerant manner, achieving approximately 74.5% classification accuracy, demonstrating that local scale invariant features can be robust even under challenging conditions.

In this letter, a method for the construction of low-level saliency maps is presented in tandem with their evaluation on a set of aerial images. One of the key inspirations for the current research lies on the observation that, usually, the most significant man-made structures in a wide-field aerial image resemble the low-level features that can be detected with a bottom-up saliency map. Aerial photography comprises, hence, a natural domain of application for a method that computationally models low-level saliency. With the employment of mechanisms analogous to the neural functions that drive human attention, we propose a bioinspired framework based on sparse coding for the extraction of information about saliency. The suggested algorithm is then evaluated on a novel data set that has been constructed with the utilization of aerial images and the corresponding manually designed ground truth binary maps of salient structures. The results demonstrate the efficiency of the proposed scheme to highlight conspicuous locations in aerial images, revealing the perspectives on the employment of low-level saliency maps in aerial imaging systems.

Autoimmune diseases are proven to be connected with the occurrence of autoantibodies in patient serum. Antinuclear autoantibodies (ANAs) identification can be accomplished in a laboratory using indirect immunofluorescence (IIF) imaging. ANAs are characterized by specific “visual” patterns on a humane epithelial cell line (HEp-2). The identification stage is usually done by trained and highly qualified physicians through visual inspection of slides using a fluorescence microscope. The presence of subjectivity in the identification process, the interobserver variability, the increasing demand of highly specialized personnel, suggest that a realization of an automatic classification system is of great significance for the field of autoimmune diseases diagnosis. Moreover CAD systems can be used in a collaborative scheme in order to augment the physicians' capabilities. In this paper a system for automatic classification of staining patterns on single-cell fluorescence images is proposed. Our method utilizes morphological features extracted from a set of binary images derived via multi-level thresholding of fluorescence images. Furthermore, a modified version of Uniform Local Binary Patterns descriptor is incorporated in order to capture local textural information. The classification is performed using a non-linear SVM Classifier. The proposed method is evaluated using a publicly available dataset, recently released for the purposes of HEP-2 Cells classification competition at ICPR 2012, achieving up to 95.9% overall classification accuracy.

In this work, a feature extraction method for off-line signature recognition and verification is proposed, described and validated. This approach is based on the exploitation of the relative pixel distribution over predetermined two and three-step paths along the signature trace. The proposed procedure can be regarded as a model for estimating the transitional probabilities of the signature stroke, arcs and angles. Partitioning the signature image with respect to its center of gravity is applied to the two-step part of the feature extraction algorithm, while an enhanced three-step algorithm utilizes the entire signature image. Fusion at feature level generates a multidimensional vector which encodes the spatial details of each writer. The classifier model is composed of the combination of a first stage similarity score along with a continuous SVM output. Results based on the estimation of the EER on domestic signature datasets and well known international corpuses demonstrate the high efficiency of the proposed methodology.

In this paper the face recognition problem is addressed in a part-based sparse approach through the comparison of respective facial regions between different images. To this purpose, a sparse coding procedure is applied to non-overlapping patches derived from frontal-face images, in order to extract local facial information. An adequate measure is introduced, incorporating the resulted sparse representation along with the Hamming distance, in order to express pairwise similarities between faces. Finally, a simple Nearest Neighbor classifier is employed to determine the identity of each facial image. In addition, a new criterion is presented for the rejection of outliers. The emerged face recognition scheme is evaluated using publicly available facial image databases, and the results are compared with those of other well-established recognition methods.

This paper proposes a novel distributional-based approach towards the face retrieval problem. Face features are initially extracted in the frequency domain via Gabor filtering, producing a large number of Gabor jets. The Neural-Gas vector quantizer is used to extract representative samples of the multivariable face distribution. In this way, only a small amount of Gabor jet signatures is utilized. Each face image is then represented as a distribution of a few signatures in the frequency space, containing all the important information. The similarity between two images is finally assessed by comparing the corresponding distributions directly in the frequency space using the multivariate Waid-Wolfowitz test (WW-test), a non- parametric statistical test dealing with the multivariate "Two-Sample Problem". Experimental results drawn from a standard collection of face-images show a significantly improved performance relative to other typical methods.

In this work, the idea of key frames extraction from single shots in video sequences is presented. The method is implemented by an efficient two-step algorithm, which is classified neither to clustering nor to temporal variations based techniques. In the first step, an MST (minimal spanning tree) graph is constructed, where each node is associated to a single frame of the shot. In the second step, extracts key frames based on the principle of their maximum spread, are extracted. The number of the selected key frames is controlled by an adaptively defined threshold, while the validity of the results is evaluated by the fidelity measure.

Locally linear embedding (LLE), isometric mapping (Isomap) are two relatively new nonlinear dimensionality reduction algorithms also used in face recognition applications. Their main aim is to create a low-dimensional embeddings of the original high-dimensional data, laying the face data points on a 'face manifold'. In this work in order to test their performance we applied LLE and Isomap in two face databases together with principal component analysis (PCA), their linear counterpart, varying as parameters the (i) number embedding dimensions and (ii) the number of neighbours. Furthermore, at the final stage we used a data ranking algorithm, which ranks the data with respect to the intrinsic manifold structure and its geometric properties. Experimental results indicate the superiority of the data ranking algorithm on face manifolds against the classical Euclidean distance measure.

A novel strategy for color-based image retrieval is introduced. Initially, a vector quantization technique is adopted, based on the application of self-organizing neural networks. The color content in each image is summarized by representative RGB-vectors extracted using the Neural-Gas network, an efficient way to extract faithful representations from multivariate distributions. The similarity between two images is then assessed as commonality between the corresponding representative color distributions and quantified using the multivariate Wald-Wolfowitz test, a nonparametric statistical test dealing with the "multivariate two-sample problem". Experimental results drawn from a diverse collection of color images show a significantly improved performance relative to the popular approach of color histogram.

In this paper, a method for content-based retrieval of JPEG images is presented, utilizing features directly from the discrete cosine transform (DCT) domain. Image indexing is achieved by extracting color and texture feature vectors, using an efficient technique applied on the DCT coefficients. Similarity between the query- and database-images is provided based on a statistical graph matching approach. The proposed measure makes use of the Wald-Wolfowitz test, a nonparametric test that assesses the commonality between two different sets of multivariate observations. Experimental results demonstrate the enhanced performance of our approach, compared to previously reported methods

In this study, a conceptually simple, yet flexible and extendable strategy to contrast two different color images is introduced. The proposed approach is based on the multivariate Wald-Wolfowitz test, a nonparametric test that assesses the commonality between two different sets of multivariate observations. It provides an aggregate gauge of the match between color images, taking into consideration all the (selected) low-level characteristics, while alleviating correspondence issues. We show that a powerful measure of similarity between two color images can emerge from the statistical comparison of their representations in a properly formed feature space. For the sake of simplicity, the RGB-space is selected as the feature space, while we are experimenting with different ways to represent the images within this space. By altering the feature-extraction implementation, complementary ways to portray the image content appear. The reported results, from the application on a diverse collection of images, clearly demonstrate the effectiveness of our method, its superiority over previous methods, and suggest that even further improvements can be achieved along the same line of research. It is not only the unifying character that makes our strategy appealing, but also the fact that the retrieval performance does not increase continuously with the amount of details in the image representation. The latter sets an upper limit to the computational demands and reminds of performance plateaus reached by novel approaches in information retrieval.

This paper proposes a methodology that incorporates principles from cluster analysis and graph representation to achieve efficient image segmentation results. More specifically, a feature-based, inter-region dissimilarity relation is considered here in order to determine the dissimilarity matrix in a graph-based segmentation scheme. The calculation of the dissimilarity function between adjacent elementary image regions is based on the proximity of each region's feature vector to the main clusters that are formed by the image samples in the feature space. In contrast to typical segmentation approaches of the literature, the global feature space information is included in the spatial graph representation that was derived from the initial Watershed partitioning. A region grouping process is applied next to form the final segmentation results. The proposed approach was also compared to approaches that use feature-based, or spatial information exclusively, to indicate its effectiveness.

A multiresolution color image segmentation approach is presented that incorporates the main principles of region-based segmentation and cluster-analysis approaches. The contribution of This work may be divided into two parts. In the first part, a multiscale dissimilarity measure is proposed that makes use of a feature transformation operation to measure the interregion relations with respect to their proximity to the main clusters of the image. As a part of this process, an original approach is also presented to generate a multiscale representation of the image information using nonparametric clustering. In the second part, a graph theoretic algorithm is proposed to synthesize regions and produce the final segmentation results. The latter algorithm emerged from a brief analysis of fuzzy similarity relations in the context of clustering algorithms. This analysis indicates that the segmentation methods in general may be formulated sufficiently and concisely by means of similarity relations theory. The proposed scheme produces satisfying results and its efficiency is indicated by comparing it with: 1) the single scale version of dissimilarity measure and 2) several earlier graph theoretic merging approaches proposed in the literature. Finally, the multiscale processing and region-synthesis properties validate our method for applications, such as object recognition, image retrieval, and emulation of human visual perception.

A multiresolution color image segmentation method is presented that incorporates the main principles of region-based and cluster analysis approaches. A multiscale dissimilarity measure in the feature space is proposed that makes use of nonparametric cluster validity analysis and fuzzy C-Means clustering. Detected clusters are utilized to assign membership functions to the image regions. In addition, a graph theoretic merging algorithm is presented that uses the formulation of fuzzy similarity relations to produce the final segmentation results. The efficiency of the resulting scheme is also experimentally indicated.

A novel computationally efficient adaptive algorithm to accomplish edge detection in multidimensional and color images is presented. It is a statistical approach, based on local, non-parametric kernel density estimation. The location of the edge discontinuity coincides with the image density function minimum and it is determined by appropriate resampling of the locally defined probability space. The operator is radially symmetric and can be easily adapted to cope with signals of any dimensionality.

In this paper a novel Fuzzy Rule Based Dissimilarity Function is presented, to determine the hierarchical merging sequence in a region based segmentation scheme. The proposed technique, based on distinct region features and fuzzy logic principles, is designed to cope with the problems inherent in the segmentation task that the traditional merging cost functions cannot overcome. It combines the global (color) and local (spatial) information of the image to compare two adjacent regions in the rgb space. The validity of the approach has been subjectively and objectively verified for several types of color images such as head and shoulders, natural and texture images.

In this paper, a new filter that performs color image enhancement is presented. The filter achieves this through the minimization of a weighted cost function. The weights are determined using potential functions which are calculated in such a way as to convey spatial information. Application of the proposed filter on a real blurred and noisy color image is performed to verify its enhancement capabilities.

A new type of edge detector is presented that uses spatial filtering of the locally thresholded image. The process is realized in two main stages. In the first one an appropriate threshold is used and the multilevel image is transformed to binary one. The second stage verifies the existence of an edge point by correlating the binary valued image points with certain binary valued spatial patterns. False edge points generated by noise are removed by means of another complementary filtering path based on the edge step size.>

Vector filtering processes are of special importance when correlated multichannel data are of interest. In this case the data sequence is taken as a whole. The idea of using vectors instead of scalers has been used extensively. This work presents a method where data dependent coefficients are derived at each time instant, upon which the filter output is computed. This nonlinear filter has properties in between the vector median and other nonlinear vector filters and similar to linear ones.>