Authorship Invarianceness For Writer Identification Using Invariant Discretization And Modified Immune Classifier

Azah Kamilah, Draman@Muda (2009) Authorship Invarianceness For Writer Identification Using Invariant Discretization And Modified Immune Classifier. PhD thesis, Universiti Teknologi Malaysia.

[img] PDF (24 Pages)
Authorship_Invarianceness_For_Writer_Identification_Using_Invariant_Discretization_And_Modified_Immune_Classifier.pdf - Submitted Version
Restricted to Registered users only

Download (3MB)


Writer Identification (WI) is an active research area in pattern recognition due to its immense potential for commercialization. It has a great significance on the criminal justice system and widely explored in forensic handwriting analysis. WI distinguishes writers based on the handwriting while ignoring the meaning of the words or written characters. The challenging process in WI is that handwriting varies from one person to another, and becomes tedious and difficult to identify the handwritten authorship accordingly. However, it is individualistic where the consistent individual features are hidden in the handwriting. Due to these, most of the previous works focus on how to acquire individual features by deploying rigid characteristics (local features). These approaches contribute to large lexicon, increase the computational complexity and decrease the accuracy. Therefore, this study focuses on extracting global features from the handwritten word shape by using the proposed United Representation technique in order to address the issues of local features in WI. The extracted features are investigated granularly to validate the existence of individual features; hence the concept of Authorship Invarianceness is introduced. Authorship Invarianceness is defined as preservation of the individual features regardless of its handwriting transformations. It improves the Authorship Invarianceness by reducing the similarity error for intra-class (same writer) and increasing the similarity error for inter-class (different writer). The individuality representation is implemented by presenting various representations of individual feature into standard representations using the proposed Invariant Discretization. Experimental results show that the performance of the proposed methods has improved with identification rate of99.90% using various classifiers including the proposed Modified Immune Classifier (MIC).

Item Type: Thesis (PhD)
Uncontrolled Keywords: Writing -- Identification, Computer vision
Subjects: H Social Sciences > H Social Sciences (General)
H Social Sciences > HV Social pathology. Social and public welfare
Divisions: Library > Disertasi > FTMK
Depositing User: Nor Aini Md. Jali
Date Deposited: 23 Nov 2014 11:45
Last Modified: 28 May 2015 04:31
Statistic Details: View Download Statistic

Actions (login required)

View Item View Item