Knowledge based Inter-point Braille document Analysis and Recognition

Abstract — Braille is a writing system used by visually impaired community. The Braille writing is embossing dots arranged in different combinations of three rows and two columns. Braille can be embossed only one side using Braille slate. Braille printer or Braille embosser can print simultaneously on both sides of a paper. Inter-point Braille document looks like having little mountains (front side embossing-recto dots) and valleys (back side embossing-verso dots). The reading finger of visually impaired will read only the bulged mountains of the Braille dots and hence only one side of the document can be read by human reading. The valleys are to be read from other side of the document. When you turn the page to read these valleys they look like mountains on that side. For human reading by visually impaired they are read separately. But when the inter-point Braille is digitized, we can recognize both sides in one scan using image processing technique. This paper presents an algorithm which analysis and recognizes both front and back embossed dots of single scanned image of an Inter-point Braille document. Algorithm uses the knowledge of Braille dot size, and dimension of the Braille cell