Pixel plot and trace based segmentation method for bilingual handwritten scripts using feedforward neural network. Issue 7 (October 2016)

• Record Type: Journal Article
• Authors: Sharma, Manoj; Dhaka, Vijay
• Is Part Of: Neural computing & applications. Volume 27:Issue 7(2016)
• Page Start: 1817
• Page End: 1829
• Publication Date: 2016-10
• Subjects: PPTRPRT -- Trace -- Offline -- Bilingual -- Handwritten -- Segmentation; Neural networks (Computer science) -- Periodicals; Neural circuitry -- Periodicals; Artificial intelligence -- Periodicals; Neural Networks (Computer) -- Periodicals; Réseaux neuronaux (Informatique) -- Périodiques; Réseaux nerveux -- Périodiques; Intelligence artificielle -- Périodiques; 006.32
• DOI: 10.1007/s00521-015-1972-2
• Languages: English
• ISSNs: 0941-0643

Abstract

Abstract In the Indian subcontinent, a number of languages are in use, and an automatic recognition of printed and handwritten scripts facilitates number of applications such as image document sorting and penetrating online libraries of image documents. This framework proposed a bilingual (English and Hindi) character-spotting framework based on feedforward neural network which works on corpus of bilingual handwritten offline documents. The proposed Pixel Plot and Trace and Re-plot and Re-trace (PPTRPRT) framework traced the actual text region of the offline handwritten bilingual scripts and lead the process of line segmentation along with skew and de-skew operations. The findings of the iterations were adopted in pixel-space-based word segmentation, which were further used in character segmentation. Moreover, PPTRPRT performs normalization operation to incorporate all pen-breadth deviations and inscription slant. The proposed framework was state of the art, reflected clearly from the findings of the framework. The proposed framework is proficient to character segmentation and provides accuracy up to 99.78 %.