Showing papers in "Pattern Recognition in 2012"

TL;DR: This work attempts to present a unifying framework through the review and comparison of some of the most important works in the literature on dataset shift, and uses different names to refer to the same concepts.

TL;DR: The results of the analysis show that for multi-label classification the best performing methods overall are random forests of predictive clustering trees (RF-PCT) and hierarchy of multi- label classifiers (HOMER), followed by binary relevance (BR) and classifier chains (CC).

TL;DR: A hybrid model of integrating the synergy of two superior classifiers: Convolutional Neural Network (CNN) and Support Vector Machine (SVM) which have proven results in recognizing different types of patterns is presented.

TL;DR: This paper analyzes key aspects of the various AIA methods, including both feature extraction and semantic learning methods and provides a comprehensive survey on automatic image annotation.

TL;DR: The method consists of three stages: region segmentation, region description and region classification, which guarantees that if a polyp is present in the image, it will be exclusively and totally contained in a single region.

TL;DR: A survey and a comparative evaluation of recent techniques for moving cast shadow detection indicate that all shadow detection approaches make different contributions and all have individual strength and weaknesses.

TL;DR: The Minkowski metric based method is experimentally validated on datasets from the UCI Machine Learning Repository and generated sets of Gaussian clusters, and appears to be competitive in comparison with other K-Means based feature weighting algorithms.

TL;DR: Experimental results show that such a hybrid combination of the HVC structure with a hierarchical classifier significantly improves expression recognition accuracy when applied to wide-ranging databases, and is not only robust to corrupted data and missing information, but can be generalized to cross-database expression recognition.

TL;DR: The proposed method deals with the joint use of the spatial and the spectral information provided by the remote-sensing images with very high spatial resolution and is competitive with other contextual methods.

TL;DR: A Markov based approach is proposed to detect image splicing and can outperform some state-of-the-art methods, making the computational cost more manageable.

TL;DR: This paper develops a computationally attractive and effective alternative to characterize the automatically segmented ear images using a pair of log-Gabor filters and presents a completely automated approach for the robust segmentation of curved region of interest using morphological operators and Fourier descriptors.

TL;DR: DSNPE not only preserves the sparse reconstructive relationship of SNPE, but also sufficiently utilizes the global discriminant structures from the following two aspects: maximum margin criterion (MMC) is added into the objective function of DSNPE.

TL;DR: A novel algorithm, called graph dual regularization non-negative matrix factorization (DNMF), which simultaneously considers the geometric structures of both the data manifold and the feature manifold is proposed.

TL;DR: An unsupervised approach for the segmentation and classification of cervical cells and performance evaluation using two data sets show the effectiveness of the proposed approach in images having inconsistent staining, poor contrast, and overlapping cells.

TL;DR: A novel inverse random under sampling (IRUS) method is proposed for the class imbalance problem and results indicate a significant increase in performance when compared with many existing class-imbalance learning methods.

TL;DR: A new algorithm that speeds up classification and a solution to reduce the training set size with negligible effects on the accuracy of classification are proposed, therefore further increasing its efficiency.

TL;DR: A new region-based active contour model, namely local region- based Chan-Vese (LRCV) model, is proposed for image segmentation, which is much more computationally efficient and much less sensitive to the initial contour.

TL;DR: This work presents an adaptive and parameterless generalization of Otsu's method, extended using a multiscale framework, and has been applied on various datasets, including the DIBCO'09 dataset, with promising results.

TL;DR: A robust EM clustering algorithm for Gaussian mixture models is developed, first creating a new way to solve these initialization problems, and then constructing a schema to automatically obtain an optimal number of clusters.

TL;DR: A feature extraction method for texture description is developed which can be integrated with existing LBP variants such as conventional LBP, rotation invariant patterns, local patterns with anisotropic structure, completed local binary pattern (CLBP) and local ternary pattern (LTP) to derive new image features for texture classification.

TL;DR: Results of incoming target sequence validate the detection capability of the proposed TM-SCR method from dim, small targets to strong, large targets in comparison with the Top-hat method at the same rate of false alarms.

TL;DR: An insight into the newly-emerging sparse representation-based classifier (SRC) is given and reasonable supports for its effectiveness are sought and it is found that for pattern recognition tasks, L"1- Optimizer provides more classification meaningful information than L"0-optimizer does.

TL;DR: A novel statistical method to automatically select the optimal subject-specific time segment and temporal frequency band based on the mutual information between the spatial-temporal patterns from the EEG signals and the corresponding neuronal activities and its one-versus-rest multi-class extension was presented.

TL;DR: Experimental results on sport player and cell tracking studies show that the proposed Bayesian method can automatically track numerous targets, and it outperforms the state-of-the-art in terms of false positive and false negative rates as detection error measures.

TL;DR: It is argued that using structured vocabularies is capital to the success of image annotation, and contributions in the field showing how structures are introduced are surveyed.

TL;DR: This paper proposes a novel face detection method using local gradient patterns (LGP), in which each bit of the LGP is assigned the value one if the neighboring gradient of a given pixel is greater than the average of eight neighboring gradients, and 0 otherwise.

TL;DR: A two-dimensional histogram equalization (2DHE) algorithm which utilizes contextual information around each pixel to enhance the contrast of an input image and is suitable for real-time contrast enhancement applications.

TL;DR: The proposed approach investigates the complexity in R, G and B color channels to characterize a texture sample and proposes to study all channels in combination, taking into consideration the correlations between them.

TL;DR: An ensemble of filters and classifiers is described to reduce the variability of the features selected by filters in different classification domains and its adequacy was demonstrated by employing 10 microarray data sets.

TL;DR: The Quill feature is a probability distribution of the relation between the ink direction and the ink width that illustrates that ink width patterns are valuable and is already being used by domain experts using a graphical interface.