CitedEvidence
User Settings

A Hybrid Segmentation System of Offline Arabic Mathematical Expression Recognition

Khalifa Mahmoud,Bingru Yang-2011-01-01-ASME Press eBooks
6

TL;DRAbstract

The hieratical adjacent characters/symbols (C/S) — lower and upper baselines - in the Simple Arabic Mathematical Equation (SAME) are the major problems in segmentation and structure recognition on a handwritten expression in document image. In this paper we briefly described the segmentation and recognition of (SAME) structure and (C/S) in still, gray-scale images. First we apply threshold for gap detection between (SAME) parts. Secondly we count the end points for each connected component (CCs) in that part for (C/S) position determining, the expression structure decoding and define the operator location. At last we apply (SOM) recognizer on the extracted feature from CCs of (SAME) parts. The system is tested on a set of handwritten and printed expressions and achieves promising results.

Chat with Paper

AI Agents for this Paper

The hieratical adjacent characters/symbols (C/S) — lower and upper baselines - in the Simple Arabic Mathematical Equation (SAME) are the major problems in segmentation and structure recognition on a handwritten expression in document image. In this paper we briefly described the segmentation and recognition of (SAME) structure and (C/S) in still, gray-scale images. First we apply threshold for gap detection between (SAME) parts. Secondly we count the end points for each connected component (CCs) in that part for (C/S) position determining, the expression structure decoding and define the operator location. At last we apply (SOM) recognizer on the extracted feature from CCs of (SAME) parts. The system is tested on a set of handwritten and printed expressions and achieves promising results.

Keywords

SegmentationArtificial intelligenceComputer sciencePattern recognition (psychology)ArabicDecoding methodsFeature (linguistics)Expression (computer science)

Chat

Click to start Chat