Research On Feature Extraction Approaches Of Handwriting Signature Using Principal Component Analysis

Signature is a relatively stable behavioral characteristic, each person's signature has some invariance. Since signatures are of the acquisition of non-invasive, they easy to be accepted by people. Signature handwriting as a personal identification are becoming increasingly importance in current society. The identification of signature handwriting are being widely used in the areas of criminal investigation, the credit card signature, e-commerce, and other fields. If a person's signature was forged, it is very likely to have very serious economic losses and social consequences. Therefore, the signature identification has critical important value and significance.In most cases, signature images are generally static, in other words, they are often off-line signature images. Therefore, in view of the broad application of off-line signature verification, this study is based on the content offline signature as a precondition. To solve the fundamental characteristics of the defects, it is necessary to apply a variety of mutual compensation. However, there are not all options will be the measurements which are used for classifying, it is likely trigger the "dimension of the disaster" phenomenon. Therefore, in order to improve the correct rate of the identification of the signature handwriting, reduce misjudgment rate and the rate of rejection, the original signature features of the full data analysis are needed. After serious option or transform, guarantee the accuracy of classification, under the premise of removing those taxonomic capacities of the poor and a strong correlation between the characteristics thereby reducing the number of features to achieve speed up the identification of speed, efficiency identification purposes.in this paper, we focus on using principal component analysis to verify the signature handwriting. Its goal is to ensure that signature handwriting feature information lost at a minimum level, and reduce the dimensions of high-dimensional space variables. Explicitness about a certain linear variables produces a relatively small number of implicit variables, reduce the original signature of the dimension of space, and then use the implicit variable from the main features. This will not only preserve the original information of signature handwriting features, but also eliminate the variables relatedness, and simplify the analysis of complexity. New features from space with the corresponding requirements of the PCA, and eliminate some of the noise, thus effectively reducing the overall classification error rate of signature identification, and achieving better results.