Baseline Image Classification Approach Using Local Minima Selection

摘要

This paper covers the area of baseline identification, which leads to signature recognition. It addresses the usage of a proposed algorithm, which identifies the minima points of a signature to be applied in signature baseline recognition. Signature baseline is vaguely identifiable and hard to determine for its baseline form. In this study, the aim is to determine the baseline form and categorizing it into ascending, descending and normal baseline. An algorithm using local minima selection technique is proposed in solving this problem. The total of 100 acquired signatures is used to determine the baseline classification range. Identifiable minima point values are extracted using an identification algorithm to yield a distribution of data that would represent the signature baseline. Then, a linear regression formula is applied to identify the direction of the baseline. The result is then tested for its accuracy with an available 100 sample of expert verified signatures. The result shows a favorable accuracy of 76% correct baseline identification. It is hoped that the implementation of this technique would be able to give some degree of contribution in the area of signature or handwriting baseline recognition.