Kenzo Obata

Constructing a high performance signature verification system using a GA method

To realize a high-peformance automatic signature verification system, it is necessary that the selected features are potentially difficult to imitate. One of the advantages of online signature verification is that the virtual strokes which are left in the pen-up situation can be obtained. These virtual strokes can be memorized by the computer but are invisible to humans. So there is little possibility of imitating these strokes deliberately. The features included in such strokes are expected to realize a high verification performance. This paper proposes to find the optimal features for signature verification from these virtual strokes by using a genetic algorithm (GA). Experiments are carried out to show the effectiveness of the new method.

Selection of features for signature verification using the genetic algorithm

This paper presents a new method to select features of handwritten signatures using the genetic algorithm. In the proposed method, the features are determined by chromosomes. The genotypes are modified by the genetic algorithm with a local improvement mechanism. A new crossover method is also proposed to determine the number of features used for signature verification. Furthermore, the moving traces of the pen in the air above the tablet are also included in the candidate strokes. Experiments are done to show the effectiveness of the new method.

A study on signature verification using a new approach to genetic based machine learning

This paper presents a new method to find best features for signature verification. The new method uses a new coding method, a new crossover method, and a new GA method with a local improvement mechanism proposed by the authors. The new coding method is effective to absorb the intra-personal variability among true signatures. The new crossover method determines the number of partial curves chosen for the signature verification. The new GA approach is very efficient in improving the local portions of chromosomes. Experiments are done to show the effectiveness of the new method.

A study on feature extraction using a fuzzy net for off-line signature recognition

This paper presents a method of off-line signature recognition using feature strokes and a fuzzy net. Each stroke has features of signatures, and the fuzzy net proposed by the authors can extract personal characteristics from the strokes. An experiment is done to show the feasibility of the new method.