Style consistent classification of isogenous patterns

Abstract

Abstract-In many applications of pattern recognition, patterns appear together in groups (fields) that have a common origin. For example, a printed word is usually a field of character patterns printed in the same font. A common origin induces consistency of style in features measured on patterns. The features of patterns co-occurring in a field are statistically dependent because they share the same, albeit unknown, style. Style constrained classifiers achieve higher classification accuracy by modeling such dependence among patterns in a field. Effects of style consistency on the distributions of field-features (concatenation of pattern features) can be modeled by hierarchical mixtures. Each field derives from a mixture of styles, while, within a field, a pattern derives from a class-style conditional mixture of Gaussians. Based on this model, an optimal style constrained classifier processes entire fields of patterns rendered in a consistent but unknown style. In a laboratory experiment, style constrained classification reduced errors on fields of printed digits by nearly 25 percent over singlet classifiers. Longer fields favor our classification method because they furnish more information about the underlying style.

Style-Based Ballot Mark Recognition

“…When a field is style-consistent, algorithms can be more effective in recognizing the underlying symbols [10]. This fits our application; we can assume that it is rare for voters to mark their ballots using a mixed style.…”

“…For this purpose, style should be exploited, not avoided. A familiar concept, this notion has already been applied with success in the field of handwriting recognition [10].…”

Estimation, Learning, and Adaptation: Systems That Improve with Use

“…Prateek Sarkar derived algorithms for optimal classification of style-consistent fields of arbitrary length [25]. He posited that the features of each pattern, while dependent on the features of other patterns in the field because of the same-style constraint, were independent of the classes of the other patterns.…”

Adaptive and Interactive Approaches to Document Analysis

“…We term style any difference between the statistical characteristics of a group of patterns generated by a single source and the characteristics of a group of patterns generated by several sources [29,30,31]. A single-source group usually exhibits some shape consistency.…”

“…The resulting optimal classifier is known as the Discrete Style Classifier [31]. The lengthy computation (exponential with field length) can be approximated by keeping track of frequently co-occurring (same-source) shapes [45] or, more consistently, by a Style First Classifier that computes the posterior probability based on the most likely style [46].…”