Handwritten word recognition with character and inter-characterneural networks

An off-line handwritten word recognition system is described. Images of handwritten words are matched to lexicons of candidate strings. A word image is segmented into primitives. The best match between sequences of unions of primitives and a lexicon string is found using dynamic programming. Neural networks assign match scores between characters and segments. Two particularly unique features are that neural networks assign confidence that pairs of segments are compatible with character confidence assignments and that this confidence is integrated into the dynamic programming. Experimental results are provided on data from the U.S. Postal Service.     

Complete mobile email management

Years have lapsed since the inauguration of the first Wireless Application Protocol (WAP) application. One of the main reasons for the lukewarm response so far may be attributed to the lack of a “Killer” application. Many companies and developers have attempted to replicate email services (an Internet Killer appplication) within the WAP realm, but till now, no significant success has been reported. This paper will first discuss on the key functions which are lacking in the present WAP-email service, and which are arguably necessary if a much larger user base is to be attracted to these services. A new system will be proposed which can enable the user to send files residing in a remote personal computer (PC) as attachments to his emails using WAP-enabled mobile devices. The implementation of the system will be presented in details. Issue relating to the connection to various types of email servers and the security of the system will also be addressed in the paper.     

Interaction bounds in multivariable control systems

Time-domain limitations due to right half-plane zeros and poles in linear multivariable control systems are studied. Lower bounds on the interaction are derived. They show not only how the location of zeros and poles are critical in multivariable systems, but also how the zero and pole directions influence the performance. The results are illustrated on the quadruple-tank process, which is a new multivariable laboratory process.     

Hardware and software architecture for power quality analysis

This paper presents a novel concept for power quality hardware and software architecture. Software tools related to detection, classification and characterization of power quality events can be integrated into a Distribution Control Center in which the interconnection between different subsystems is being carried out through the Internet allowing flexibility to the system. This system is implemented using wavelet analysis applied in all proposed algorithms. Different power quality events are taken as examples to illustrate the capabilities of the proposed method. The obtained results reveal that the proposed architecture is feasible to be implemented as an integral part of a control center system.The proposed system has the property of fast and accurate detection and classification of any power quality disturbance event and introduces a new PQ index determination that allows characterizing any type of disturbance including the non-periodic signals.     

Mechanical properties of iron processed by severe plastic deformation

In the present study, the mechanical properties of Fe processed via severe plastic deformation (equal-channel angular pressing (ECAP)) at room temperature were investigated for the first time. The grain size of annealed Fe, with an initial grain size of about 200 μm, was reduced drastically during ECAP. After eight passes, the grain size reaches 200 to 400 nm, as documented by means of transmission electron microscopy (TEM). The value of microhardness during pressing increases 3 times over that of the starting material after the first pass and increases slightly during subsequent pressing for higher-purity Fe. Examination of the value of microhardness after eight passes as a function of post-ECAP annealing temperature shows a transition from recovery to recrystallization, an observation that resembles the behavior reported for heavily deformed metals and alloys. The tensile and compression behaviors were examined. In tension, a drop in the engineering stress-engineering strain curve beyond maximum load was observed both in the annealed Fe and the ECAP Fe. This drop is related to the neck deformation. The fracture surface, examined by scanning electron microscopy (SEM), shows vein patterns, which is different from the dimples found on the fracture surface of annealed Fe. In compression, an initial strain-hardening region followed by a no-strain-hardening region was observed in the ECAP Fe. The yield strength in tension of the ECAP Fe was observed to be higher than that in compression. The strengthening mechanisms and softening behavior are discussed.     

Extensions of LDA by PCA mixture model and class-wise features

Linear discriminant analysis (LDA) is a data discrimination technique that seeks transformation to maximize the ratio of the between-class scatter and the within-class scatter. While it has been successfully applied to several applications, it has two limitations, both concerning the underfitting problem. First, it fails to discriminate data with complex distributions since all data in each class are assumed to be distributed in the Gaussian manner. Second, it can lose class-wise information, since it produces only one transformation over the entire range of classes. We propose three extensions of LDA to overcome the above problems. The first extension overcomes the first problem by modelling the within-class scatter using a PCA mixture model that can represent more complex distribution. The second extension overcomes the second problem by taking different transformation for each class in order to provide class-wise features. The third extension combines these two modifications by representing each class in terms of the PCA mixture model and taking different transformation for each mixture component. It is shown that all our proposed extensions of LDA outperform LDA concerning classification errors for synthetic data classification, hand-written digit recognition, and alphabet recognition.     

A tolerant approach to reconstruct topology from unorganized trimmed surfaces

Presented in this paper is a topology reconstruction algorithm from a set of unorganized trimmed surfaces. Error-prone small geometric elements are handled to give proper topological information. It gives complete topology to topologically complete models, and it is also tolerant to incomplete models. The proposed algorithm is vertex-based in that clues for topological information are searched from the set of vertices first, not from that of edges.