Showing papers by "Ching Y. Suen published in 2014"

Investigating of Preprocessing Techniques and Novel Features in Recognition of Handwritten Arabic Characters

Abstract: There are many difficulties facing a handwritten Arabic recognition system such as unlimited variation in character shapes. This paper describes a new method for handwritten Arabic character recognition. We propose a novel efficient approach for the recognition of off-line Arabic handwritten characters. The approach is based on novel preprocessing operations, structural statistical and topological features from the main body of the character and also from the secondary components. Evaluation of the importance and accuracy of the selected features was made. Our method based on the selected features and the system was built, trained and tested by CENPRMI dataset. We used SVM (RBF) and KNN for classification to find the recognition accuracy. The proposed algorithm obtained promising results in terms of accuracy; with recognition rates of 89.2% for SVM. Compared with other related works and also our recently published work we find that our result is the highest among them.

Artificial Neural Networks in Pattern Recognition: 6th IAPR TC 3 International Workshop, ANNPR 2014, Montreal, QC, Canada, October 6-8, 2014

Abstract: This book constitutes the refereed proceedings of the 6th IAPR TC3 International Workshop on Artificial Neural Networks in Pattern Recognition, ANNPR 2014, held in Montreal, QC, Canada, in October 2014. The 24 revised full papers presented were carefully reviewed and selected from 37 submissions for inclusion in this volume. They cover a large range of topics in the field of learning algorithms and architectures and discussing the latest research, results, and ideas in these areas.

Novel Global and Local Features for Near-Duplicate Document Image Matching

Abstract: A new near-duplicate document image matching approach is proposed. Globally, we model the spatial arrangements of objects in an image. Locally, the micro-patterns within each object are captured. To define a micro-pattern, the VV-nary center-symmetric gray value differences in an image local neighborhood of a variable radius are exploited. A visual descriptor is proposed to characterize the appearance of the object based on micro-pattern distributions. By combining the global and local features, each document image is represented by a compact signature with a variable length. We employ Earth Mover's Distance for image dissimilarity computation, which stands out for its remarkable ability to tolerate the instability of object segmentation by allowing many-to-many correspondence among objects. Extensive experiments on two data sets demonstrate the effectiveness of the proposed approach.

End-Shape Recognition for Arabic Handwritten Text Segmentation

Abstract: Text segmentation is an essential pre-processing stage for many systems such as text recognition and word spotting. However, few methods have been published for Arabic text segmentation. In Arabic handwritten documents, separating text into words is challenging due to the enormous different Arabic handwriting styles. In this paper, we present a new segmentation methodology of an Arabic handwritten text line into words. Our proposed approach of text segmentation utilizes the knowledge of Arabic writing characteristics. This method shows promising results.

Part-Based High Accuracy Recognition of Serial Numbers in Bank Notes

Abstract: This paper proposes a novel part-based character recognition method for a new topic of RMB (renminbi bank note, the paper currency used in China) serial number recognition, which is important for reducing financial crime and improving financial market stability and social security. Given an input sample, we first generate a set of local image parts using the Difference-of-Gaussians (DoG) keypoint detector. Then, all of the local parts are classified by an SVM classifier to provide a confidence vector for each part. Finally, three methods are introduced to combine the recognition results of all parts. Since the serial number samples suffer from complex background, occlusion, and degradation, our part-based method takes advantage of both global and local character structure features, and offers an overall increase in robustness and reliability to the entire recognition system. Experiments conducted on a RMB serial number character database show that the test accuracy boosted from 98.90% to 99.33% by utilizing the proposed method with multiple voting based combination strategy. The part-based recognition method can also be extended to other types of banknotes, such as Euro, U.S. and Canadian dollars, or in character recognition applications with complex backgrounds.

Comic2CEBX: a system for automatic comic content adaptation

Abstract: Comics are popular almost throughout the world. With the help of comic document digitization, it is much easier for people to archive and browse comic works. However, there are still some big challenges along with comic document digitization progress. Among these challenges, comic content adaptation is an important one to be tackled. The existing works only focus on parts of this problem and do not provide a tangible solution to display comic contents on different devices. In this paper, we solve these problems by proposing Comic2CEBX, a system which can automatically convert a set of scanned comic page images into a CEBX file that allows reflowing of the original comic pages with fixed layouts. Taking raw comic images as inputs, our system first extracts three kinds of low-level visual patterns and then uses multilayer Conditional Random Fields to detect all the panels. Meanwhile, our system automatically identifies the reading orders of the panels within each page. Finally, we encapsulate the comic page images and the obtained page structure information (i.e., the panels detection results and the corresponding reading orders) to generate a CEBX file. Experimental results show that our comic page layout analysis method achieves better performance than the existing ones, and use case presentation of the CEBX files produced by our system demonstrates that it brings better comic reading experience especially on mobile devices.

The Effect of Correlation and Performances of Base-Experts on Score Fusion

Abstract: In the field of biometric authentication, it is a promising trend to perform score fusion to improve authentication accuracy. Many empirical studies have shown the effectiveness of score fusion; however, some other researchers assert that fusion is not always beneficial. Despite considerable empirical efforts, to the best of our knowledge, the research devoted to the theoretical analysis of fusion can be found only in the paper by Poh and Bengio published in 2005. Unfortunately, we find that the variance reduction-equal error rate (VR-EER) model, which is the theoretical basis of this reference, is incorrect and the resulting conclusions are arguable. Besides, we find that the conclusions from several other empirical studies are arguable too. In this paper, using Fermat's theorem and the connection between F-ratio and EER, we conduct a systematic theoretical study on how correlation and performances of base-experts affect fusion, giving the underlying reason why VR-EER model and the above conclusions are wrong. Contrary to these existing conclusions, we prove that provided fusion weights are selected according to our proposed criterion, the combined system will definitely be superior to all the base-experts, regardless of correlation, performances, or variances of base-experts. Experiments are carried out to validate the conclusions of ours and construct counter-examples for the existing conclusions.

Novel Handwritten Words and Documents Databases of Five Middle Eastern Languages

Abstract: This paper introduces new handwritten databases of selected words in the five Middle-Eastern languages of Arabic, Dari, Farsi, Pashto and Urdu. The databases share a common lexicon of forty words that are related to finance and are used in daily life. The five databases have been collected from over 1600 native writers located in four countries. Recognition results for each of the databases are also presented. Results come from three classifiers (Support Vector Machines, Modified Quadratic Discriminant Function. And Multi-layer Perceptron) which were implemented for recognition of the words based on gradient features. Given the diversity of the data, the results demonstrate the effectiveness of the implemented process in learning and recognizing samples of handwritten words from different languages. In addition, full page handwritten documents of each language are presented, with approximately forty pages per language. Each document has associated ground truth information.

Application of PR Techniques to Mail Sorting in China

Abstract: Mail sorting machines play an important role in postal automation. In this paper, we give a brief overview of mail sorting machines in China Post from a pattern recognition point of view. OCR techniques such as postcode recognition and address recognition are essential for mail sorting machines, which are considered as class imbalance problems in our study. In addition to OCR that mainly focuses on the characters in the image, some image-level features are exploited for near-duplicate envelope image matching, which is helpful in tracking pieces of mail in the network environment.

Face Recognition Based on Discriminative Dictionary with Multilevel Feature Fusion

Abstract: In order to alleviate the influence of illumination, pose, expression and occlusion variations in face recognition, in this paper, an effective face recognition method based on discriminative sparse representation is proposed. To solve the problem of these variations, we extract discriminative features which represent for each of the training images, and propose a novel dictionary by learning discriminative features. Firstly, we decompose a test image by using nonsubsampled contourlet transform (NSCT), and then fuse the information according to the features from each subband and their contributions. Finally, we obtain the discriminative features of training images and construct a discriminative dictionary. Fuse these multiple features can improve the efficiency and effectiveness of face recognition, especially when training samples are limited and the dimension of feature vector is low. Experimental results on two widely used face databases are presented to demonstrate the efficiency of the proposed approach.

Convex Cardinality Restricted Boltzmann Machine and Its Application to Pattern Recognition

Abstract: The Restricted Boltzmann machine is a graphical model which has been very successful in machine learning and various applications. Recently lots of attention has been devoted to sparse techniques combining a cardinality potential function and an energy function. In this paper we use a convex cardinality potential function for increasing competition between hidden units to be sparser. Convex potential functions relax conditions on hidden units, encouraging sparsity among them. In this paper we show that combination of convex potential function with cardinality potential produces better results in classification.