Writer Identification Using Handwritten Cursive Texts and Single Character Words

Abstract

One of the biometric methods in authentication systems is the writer verification/identification using password handwriting. The main objective of this paper is to present a robust writer verification system by using cursive texts as well as block letter words. To evaluate the system, two datasets have been used. One of them is called Secure Password DB 150, which is composed of 150 users with 18 samples of single character words per user. Another dataset is public and called IAM online handwriting database, and it is composed of 220 users of cursive text samples. Each sample has been defined by a set of features, composed of 67 geometrical, statistical, and temporal features. In order to get more discriminative information, two feature reduction methods have been applied, Fisher Score and Info Gain Attribute Evaluation. Finally, the classification system has been implemented by hold-out cross validation and k-folds cross validation strategies for three different classifiers, K-NN, Naïve Bayes and Bayes Net classifiers. Besides, it has been applied for verification and identification approaches. The best results of 95.38% correct classification are achieved by using the k-nearest neighbor classifier for single character DB. A feature reduction by Info Gain Attribute Evaluation improves the results for Naïve Bayes Classifier to 98.34% for IAM online handwriting DB. It is concluded that the set of features and its reduction are a strong selection for the based-password handwritten writer identification in comparison with the state-of-the-art.

Writer Recognition Using Off-line Handwritten Single Block Characters

“…Some related works [7,8] have a similar purpose, but using passwords written on a smartphone touch-screen, thus accounting for on-line information. In [7], the authors use passwords of 8 characters. The best reported identification accuracy is 95.38% with a combination of geometrical, statistical and temporal features.…”

A Performance of Combined Methods of VCG and 16BCD for Feature Extraction on HSV

“…Reference [6] proposed an algorithm to correct the alignment of the input signature which can be used at the pre-processing stage to achieve better results in the signature detection process. Several methods of extracting statistical features have also discussed in [7][8][9][10][11][12]. Reference [13] had used Gabor Filter to extract the surf feature and critical point matching, while the classification is based on Hidden Markov Model (HMM).…”

Sequential Authentication Using Handwriting Biometrics for Free Response e-Testing

“…a) E-mail: 4417701@ed.tus.ac.jp b) E-mail: akakura@rs.tus.ac.jp DOI: 10.1587/transinf.2019EDL8144 mance of the previous identification system which neglects the dynamic information between observations. Kutzner et al [5] proposed writer authentication using various geometrical, statistical and dynamic features in handwritten cursive texts and single character words. These method [4], [5] has high accuracy with the dynamic information of pen coordinate, but we assumed the pen angle can be taken for prevent to forged writing such as a text made by tracing or copy.…”

“…Kutzner et al [5] proposed writer authentication using various geometrical, statistical and dynamic features in handwritten cursive texts and single character words. These method [4], [5] has high accuracy with the dynamic information of pen coordinate, but we assumed the pen angle can be taken for prevent to forged writing such as a text made by tracing or copy. In addition, these researches [3]- [5] assume to identify users at the end of the character or text input and calculate the likelihood of being a principle from all of a subdivided character such as stroke unit.…”

“…These method [4], [5] has high accuracy with the dynamic information of pen coordinate, but we assumed the pen angle can be taken for prevent to forged writing such as a text made by tracing or copy. In addition, these researches [3]- [5] assume to identify users at the end of the character or text input and calculate the likelihood of being a principle from all of a subdivided character such as stroke unit. So, it is difficult to apply to the free response tests that have to authenticate the examinee in the middle of writing an answer.…”