Off-Line Writer Verification: A Comparison of a Hidden Markov Model (HMM) and a Gaussian Mixture Model (GMM) Based System
Abstract
In this paper, we introduce and compare two off-line, text independent writer verification systems. At the core of the first system are Hidden Markov Model (HMM) based recognizers. The second system uses Gaussian Mixture Models (GMMs) to model a person's handwriting. Both systems are evaluated on two test sets consisting of unskillfully forged and skillfully forged text lines, respectively. In this comparison, different confidence measures are considered, based on the raw log-likelihood score, the cohort model approach, and the world model approach.
Loading...