International Journal of Pattern Recognition and Artificial Intelligence

In this paper, we present a pattern recognition method that uses dynamic programming for the alignment of Radon features. The key characteristic of the method is to use dynamic time warping (DTW) to match corresponding pairs of the Radon features for all possible projections. Thanks to DTW, we avoid compressing the feature matrix into a single vector which would otherwise miss information. To reduce the possible number of matchings, we rely on a initial normalization based on the pattern orientation. A comprehensive study is made using major state-of-the-art shape descriptors over several public datasets of shapes such as graphical symbols (both printed and hand-drawn), handwritten characters and footwear prints. In all tests, the method proves its generic behavior by providing better recognition performance. Overall, we validate that our method is robust to deformed shape due to distortion, degradation and occlusion.

In this paper, we propose a method for text-query-based keyword spotting from online Chinese handwritten documents using character classification model. The similarity between the query word and handwriting is obtained by combining the character classification scores. The classifier is trained by one-versus-all strategy so that it gives high similarity to the target class and low scores to the others. Using character classification-based word similarity also helps overcome the out-of-vocabulary (OOV) problem. We use a character-synchronous dynamic search algorithm to efficiently spot the query word in large database. The retrieval performance is further improved by using competing character confusion and writer-adaptive thresholds. Our experimental results on a large handwriting database CASIA-OLHWDB justify the superiority of one-versus-all trained classifiers and the benefits of confidence transformation, character confusion and adaptive thresholds. Particularly, a one-versus-all trained prototype classifier performs as well as a linear support vector machine (SVM) classifier, but consumes much less storage of index file. The experimental comparison with keyword spotting based on handwritten text recognition also demonstrates the effectiveness of the proposed method.

Vision-based robot localization outdoors has remained more elusive than its indoors counterpart. Drastic illumination changes and the scarceness of suitable landmarks are the main difficulties. This paper attempts to surmount them by deviating from the main trend of using local features. Instead, a global descriptor called landmark-view is defined, which aggregates the most visually-salient landmarks present in each scene. Thus, landmark co-occurrence and spatial and saliency relationships between them are added to the single landmark characterization, based on saliency and color distribution. A suitable framework to compare landmark-views is developed, and it is shown how this remarkably enhances the recognition performance, compared against single landmark recognition. A view-matching model is constructed using logistic regression. Experimentation using 45 views, acquired outdoors, containing 273 landmarks, yielded good recognition results. The overall percentage of correct view classification obtained was 80.6%, indicating the adequacy of the approach.

In this study, a fuzzy clustering algorithm, MRHMRF-FCM, is proposed to capture and utilize the multiscale spatial constrains by employing multiresolution representations for the label image and the observed image in wavelet domain. In this algorithm, the inner-scale and inter-scale spatial constrains, respectively modeled by the hidden Markov random field models, serve as the penalization terms for the objective function of the FCM algorithm. On each scale, the improved objective function is optimized by taking advantage of Lagrange multipliers, and the final label of wavelet coefficient is determined by iteratively updating the membership degree and cluster centers. The experimental results on synthetic images, natural scenery color images and remote sensed images show that the proposed algorithm obtains much better segmentation results, such as accurately differentiating different regions and being immune to noise.

In this paper, a new approach based on score level fusion is presented to obtain a robust recognition system by concatenating face and iris scores of several standard classifiers. The proposed method concatenates face and iris match scores instead of concatenating features as in feature-level fusion. The features from face and iris are extracted using local and global feature extraction methods such as PCA, subspace LDA, spPCA, mPCA and LBP. Transformation-based score fusion and classifier-based score fusion are then involved in the process to obtain, concatenate and classify the matching scores. Different fusion techniques at matching score level, feature level and decision level are compared with the proposed method to emphasize improvement and effectiveness of the proposed method. In order to validate the proposed scheme, a combined database is formed using ORL and BANCA face databases together with CASIA and UBIRIS iris databases. The results based on recognition performance and ROC analysis demonstrate that the proposed score level fusion achieves a significant improvement over unimodal methods and other multimodal face-iris fusion methods.

Graph-based representations of patterns are very flexible and powerful, but they are not easily processed due to the lack of learning algorithms in the domain of graphs. Embedding a graph into a vector space solves this problem since graphs are turned into feature vectors and thus all the statistical learning machinery becomes available for graph input patterns. In this work we present a new way of embedding discrete attributed graphs into vector spaces using node and edge label frequencies. The methodology is experimentally tested on graph classification problems, using patterns of different nature, and it is shown to be competitive to state-of-the-art classification algorithms for graphs, while being computationally much more efficient.

Training a support vector machine (SVM) requires the solution of a quadratic programming problem (QP) whose computational complexity becomes prohibitively expensive for large scale datasets. Traditional optimization methods cannot be directly applied in these cases, mainly due to memory restrictions. By adopting a slightly different objective function and under mild conditions on the kernel used within the model, efficient algorithms to train SVMs have been devised under the name of core vector machines (CVMs). This framework exploits the equivalence of the resulting learning problem with the task of building a minimal enclosing ball (MEB) problem in a feature space, where data is implicitly embedded by a kernel function. In this paper, we improve on the CVM approach by proposing two novel methods to build SVMs based on the Frank–Wolfe algorithm, recently revisited as a fast method to approximate the solution of a MEB problem. In contrast to CVMs, our algorithms do not require to compute the solutions of a sequence of increasingly complex QPs and are defined by using only analytic optimization steps. Experiments on a large collection of datasets show that our methods scale better than CVMs in most cases, sometimes at the price of a slightly lower accuracy. As CVMs, the proposed methods can be easily extended to machine learning problems other than binary classification. However, effective classifiers are also obtained using kernels which do not satisfy the condition required by CVMs, and thus our methods can be used for a wider set of problems.

The performance of face recognition systems tends to decrease when images are affected by illumination. Feature extraction is one of the main steps of a face recognition process, where it is possible to alleviate the illumination effects on face images. In order to increase the accuracy of recognition tasks, different methods for obtaining illumination invariant features have been developed. The aim of this work is to compare two different ways to represent face image descriptions in terms of their illumination invariant properties for face recognition. The first representation is constructed following the structure of complex numbers and the second one is based on quaternion numbers. Using four different face description approaches both representations are constructed, transformed into frequency domain and expressed in polar coordinates. The most illumination invariant component of each frequency domain representation is determined and used as the representative information of the face image. Verification and identification experiments are then performed in order to compare the discriminative power of the selected components. Representative component of the quaternion representation overcame the complex one.

Variations in illumination is one of major limiting factors of face recognition system performance. The effect of changes in the incident light on face images is analyzed, as well as its influence on the low frequency components of the image. Starting from this analysis, a new photometric normalization method for illumination invariant face recognition is presented. Low-frequency Discrete Cosine Transform coefficients in the logarithmic domain are used in a local way to reconstruct a slowly varying component of the face image which is caused by illumination. After smoothing, this component is subtracted from the original logarithmic image to compensate for illumination variations. Compared to other preprocessing algorithms, our method achieved a very good performance with a total error rate very similar to that produced by the best performing state-of-the-art algorithm. An in-depth analysis of the two preprocessing methods revealed notable differences in their behavior, which is exploited in a multiple classifier fusion framework to achieve further performance improvement. The superiority of the proposal is demonstrated in both face verification and identification experiments.

A technique for color quantization is described, which consists of two processes. The first process is based on the analysis of the histograms of the three color components of the RGB input image. The second process performs clustering of the colors quantized by the first process, based on their Euclidean distance. At the end of the second process, the output image is obtained by replacing the color of each pixel of the input image with the closest representative color. The obtained results are satisfactory from both the qualitative and the quantitative point of view.