International Journal of Pattern Recognition and Artificial Intelligence

There are different kinds of multiple classifiers in complex recognition systems in pursuit of better recognition capabilities. To exploit the classifiers' potential as individual ones sufficiently and enable them to work cooperatively for the best classification results, they need to be considered as a whole and be dynamically managed according to the changing recognition occasions. In this paper, we present the conception of distributed Multiple Classifiers Management (MCM) and a self-adaptive recursive MCM model based on Mixture-of-Experts (ME). A control subsystem is consisted in the model, which allows the classification progress to be controlled by the systems' priori information when necessary. The model adjusts its parameters dynamically according to the current recognition state and gives the recognition results by combining the current individual classifiers' results with the previous combination result under priori information's control. An algorithm based on one step error correction is presented to acquire the model's parameters dynamically. It takes the previous times' ensemble classification results as true and corrects the current weights of the classifiers. At last, an experiment on the recognition of space objects is simulated. The experiment results show that the MCM model in this paper is effective for complex recognition system containing heterogeneous classifiers on improving the recognition rate and robustness.

Missing data approaches have recently been applied to speech recognition tasks to increase noise robustness. The drawback of missing data techniques is the vulnerability of the recognizer to errors in the reliability mask. This work proposes a novel group selection algorithm to perform top-down refinement of initial bottom-up reliability mask estimates with the goal of removing these errors. A novel probabilistic decision process based on normalized likelihood distances is proposed and used to evaluate the quality of a reliability mask without any a priori noise knowledge. Experimental results on a speaker identification task illustrate the ability of the combined bottom-up top-down system to significantly outperform traditional bottom-up only missing data techniques for various types of mask corruption.

For the importance of communication between human and machine interface, it would be valuable to develop an implement which has the ability to recognize emotional states. In this paper, we proposed an approach which can deal with the daily dependence and personal dependence in the data of multiple subjects and samples. 30 features were extracted from the physiological signals of subject for three states of emotion. The physiological signals measured were: electrocardiogram (ECG), skin temperature (SKT) and galvanic skin response (GSR). After removing the daily dependence and personal dependence by the statistical technique of MANOVA, six machine learning methods including Bayesian network learning, naive Bayesian classification, SVM, decision tree of C4.5, Logistic model and K-nearest-neighbor (KNN) were implemented to differentiate the emotional states. The results showed that Logistic model gives the best classification accuracy and the statistical technique of MANOVA can significantly improve the performance of all six machine learning methods in emotion recognition system.

Similarity measurement is crucial for classification. Based on the manifold assumption, many graph-based algorithms were developed. Almost all methods follow the k-rule or ε-rule to construct a graph, and then focus on the algorithms based on the graph. However, the graph may not represent the local structure well, and it does not fully utilize the label information yet. The local structure can be presented by the local density and the distance between the samples and their neighbors. And the graph constructed by the guidance of label information will be better approximate of the relationship of the input data. In this paper, we propose an adaptive semi-supervised graph constructing method. The similarity is learned when constructing the graph. The advantages of the similarity learned by our method include: (1) The similarity is measured along the manifold by constructing a graph; (2) nearby points and points in the same cluster share high similarity; (3) samples from the same class have higher similarity than samples from different classes. Experimental results show that using the proposed similarity for classification task could get better recognition accuracy.

Associating features with weights is a common approach in clustering algorithms and determining the weight values is crucial in generating valid partitions. In this paper, we introduce a novel method in the framework of granular computing that incorporates fuzzy sets, rough sets and shadowed sets, and calculates feature weights automatically. Experiments on synthetic and real data patterns show that our algorithms always converge and are more effective in handling overlapping among clusters and more robust in the presence of noisy data and outliers.

In document image analysis (DIA) especially in handwritten document recognition, standard databases play significant roles for evaluating performances of algorithms and comparing results obtained by different groups of researchers. The field of DIA regard to Indo-Persian documents is still at its infancy compared to Latin script-based documents; as such standard datasets are not still available in literature. This paper is an effort towards alleviating this gap. In this paper, an unconstrained handwritten dataset containing documents of Persian, Bangla, Oriya and Kannada (PBOK) is introduced. The PBOK contains 707 text-pages written in four different languages (Persian, Bangla, Oriya and Kannada) by 436 individuals. Total number of text-lines, words/subwords and characters are 12,565, 104,541 and 423,980, respectively. In most documents of PBOK dataset contain either an overlapping or a touching text-lines. The average number of text-lines in text-pages of the PBOK dataset is 18. Two types of ground truths, based on pixels information and content information, are generated for the dataset. Because of such ground truths, the PBOK dataset can be utilized in many areas of document image processing e.g. text-line segmentation, word segmentation and word recognition. To provide an insight for other researches, recent text-line segmentation results on this dataset are also reported.

Locating text region from an image of nature scene is significantly helpful for better understanding the semantic meaning of the image, which plays an important role in many applications such as image retrieval, image categorization, social media processing, etc. Traditional approach relies on the low level image features to progressively locate the candidate text regions. However, these approaches often suffer for the cases of the clutter background since the adopted low level image features are fairly simple which may not reliably distinguish text region from the clutter background. Motivated by the recent popular research on attention model, salience detection is revisited in this paper. Based on the case of text detection on nature scene image, saliency map is further analyzed and is adjusted accordingly. Using the adjusted saliency map, the candidate text regions detected by the common low level features are further verified. Moreover, efficient low level text feature, Histogram of Edge-direction (HOE), is adopted in this paper, which statistically describes the edge direction information of the region of interest on the image. Encouraging experimental results have been obtained on the nature scene images with the text of various languages.

Automatic fingerprint identification systems (AFIS) have been studied extensively and are widely used for biometric identification. Given its importance, many well-engineered methods have been developed for the different stages that encompass those systems. The first stage of any such system is the segmentation of the actual fingerprint region from the background. This is typically achieved by classifying pixels, or blocks of pixels, based on a set of features. In this paper, we describe novel features for fingerprint segmentation that express the underlying manifold topology associated with image patches in a local neighborhood. It is shown that fingerprint patches seen in a high-dimensional space form a simple and highly regular circular manifold. The characterization of the manifold topology suggests a set of optimal features that characterize the local properties of the fingerprint. Thus, fingerprint segmentation can be formulated as a classification problem based on the deviation from the expected topology. This leads to features that are more robust to changes in contrast than mean, variance and coherence. The superior performance of the proposed features for fingerprint segmentation is shown in the eight datasets from the 2002 and 2004 Fingerprint Verification Competitions.

Nowadays many techniques are being used to increase the reliability of human identification systems. Iris is a part of human body that is desirable for biometric identification and has favorable factors. We have focused on the reality that iris is a fractal phenomenon in this paper. During the production of new fractals, some features will be extracted by Chaos Game mechanism. These features are useful and effective in iris identification. There are three steps for iris identification with fractal and Chaos Game Theory. The first step is making a new fractal. The second step includes extracting features during the first step. Finally, the iris identification based on extracted features is the third step. We have named this technique Iris Identification based-Fractal and Chaos Game Theory (Iris-IFCGT). This technique has some fractal properties like stability against zoom, removing part of the iris image, no sensitivity on rotation and so on as well as desirable speed which helps preventing time consuming process of pattern recognition.

To design an effective and robust fingerprint recognition method is still an open issue. Some texture-based methods such as directional energy method and wavelet method are available nowadays. However, directional energy method is insufficient to capture the detail information of fingerprint and it is also improper to directly use wavelet method to extract the feature since the complex and rich edge information of fingerprint. In this work we propose a texture-based method called DFB-Wavelet for fingerprint recognition via combining directional filter banks (DFB) and wavelet. The region of interest (ROI) composed of nonoverlapping square blocks, is decomposed into eight directions by employing DFB. Wavelet signatures are calculated as the features of a fingerprint image from each directional subband of DFB. The feature matching is performed on the global normalized Euclidean distance between the input fingerprint features and the templates features. Experimental results show that DFB-Wavelet method has the higher accuracy compared to the traditional texture-based methods.