In this paper, we consider the challenging problem of unusual event detection in video surveillance systems. The proposed approach makes a step toward generic and automatic detection of unusual events in terms of velocity and acceleration. At first, the moving objects in the scene are detected and tracked. A better representation of moving objects trajectories is then achieved by means of appropriate pre-processing techniques. A supervised support vector machine method is then used to train the system with one or more typical sequences, and the resulting model is then used for testing the proposed method with other typical sequences (different scenes and scenarios). Experimental results are shown to be promising. The presented approach is capable of determining similar unusual events as in the training sequences.

On-line recognition differs from off-line recognition in that additional information about the drawing order of the strokes is available. This temporal information makes it easier to recognize handwritten texts with an on-line recognition system. In this paper we present a method for the recovery of the stroke order from static handwritten images. The algorithm was tested by classifying the words of an off-line database with a state-of-the-art on-line recognition system. On this database with 150 different words, written by four cooperative writers, a recognition rate of 97.4% was obtained.

Building large optical character recognition (OCR) databases is time-consuming and tedious. Moreover, the process is error-prone due to the difficulty in segmentation and the uncertainty in labelling. When the database is very large, say one million patterns, human errors due to fatigue and inattention become a critical factor. This paper discusses one method to alleviate the burden caused by these problems. Specifically, the method allows an automatic detection of abnormalities, e.g. mislabelling, and thus may contribute to clean up a labelled database. The method is based on the optimum class-selective rejection rule. As a test case, the method is applied to the NIST databases containing nearly 300,000 handwritten numerals.

This paper discusses several aspects of automatic handwriting recognition designed for a postal check reading system. We first present the general methodology, including a new approach called perturbation method. Then we explain how the general methodology is applied to three subproblems in handwriting recognition, namely, the recognition of isolated numerals: that of numeral strings, and the recognition of cursively handwritten words drawn from a small lexicon. Experimental results performed on worldwide standard databases show that our systems are either equivalent to or better than state-of-the-art systems. A preliminary experiment of reading real Swiss checks shows that the system rejects about 50% and correctly recognises the remaining checks.

Class-selective rejection is an extension of simple rejection. That is, when an input pattern cannot be reliably assigned to one of the N classes in an N-class problem, it is assigned to a subset of classes that are most likely to issue the pattern, instead of simply being rejected. By selecting more classes, the risk of making an error can be reduced, at the price of subsequently having a larger remaining number of classes. The optimality of class-selective rejection is therefore defined as the best trade-off between error rate and average number of selected classes. Formally, the trade-off study is embedded in the framework of decision theory. The average expected loss is expressed as a linear combination of error rate and average number of classes. The minimization of the average expected loss, therefore, provides the best trade-off. The complexity of the resulting optimum rule is reduced, via a discrete convex minimization, to be linear in the number of classes. Upper-bounds on error rate and average number of classes are derived. An example is provided to illustrate various aspects of the optimum decision rule. Finally, the implications of the new decision rule are discussed.

This paper presents a new approach to off-line, handwritten numeral recognition. From the concept of perturbation due to writing habits and instruments, we propose a recognition method which is able to account for a variety of distortions due to eccentric handwriting. We tested our method on two worldwide standard databases of isolated numerals, namely CEDAR and NIST, and obtained 99.09 percent and 99.54 percent correct recognition rates at no-rejection level respectively. The latter result was obtained by testing on more than 170000 numerals.

The concept of rejection is extended to that of class-selective rejection. That is, when an input pattern cannot be reliably assigned to one of the N classes in a N-class problem, it is assigned to a subset of classes that are most likely to issue the pattern, instead of simply being rejected. First, a new optimality criterion is appropriately defined to accommodate the newly introduced decision outcomes. Then, a new decision rule is proposed and its optimality proven. Various upper-bounds of error rate are obtained. Next, an example is provided to illustrate various aspects of the optimum decision rule. Finally, the implications of the new decision rule are discussed.

We propose a system for segmentation and recognition of totally unconstrained handwritten numeral strings. The system is built upon a number of components, namely, a presegmentation module, an isolated numeral recognizer, a segmentation-free module and a merging module. Presegmentation consists in dividing the input numeral string image into groups of numerals each of which represents an integer number of numerals. For each group, the actual number of numerals and their identity are then determined by a cascade of two recognition-based tests: isolated numeral and segmentation-free. The last one is able to recognise a numeral group of any length. All results from all groups are eventually merged yielding the final interpretation of the input numeral string. We also introduce the concept of dummy symbol in order to overcome the problem of noisy parts that cannot be eliminated by standard filtering algorithms. Preliminary experiments on totally unconstrained data from the CEDAR database yielded a segmentation rate of 94.5% and a recognition rate of 84.2% at string level. These results compare favorably to other published methods.

Check forms are used by many people in daily life for money remittances. Surprisingly, the processing of these forms at banks and post offices is only partly automated. In this paper, we consider a particular kind of form, viz., the GIRO check forms used in Switzerland. We will describe a fully automatic system which is able to recognise the financial institution, the name and address of the receiver, and the account number on a GIRO check. The system comprises procedures for binarization, segmentation, model matching, and optical character recognition (OCR). Experiments on a sample set of 48 checks have shown promising results in terms of both computation time and recognition accuracy.