Text-Line Extraction for Arabic Handwritten Documents

n this project, we used classical computer vision techniques to extract text lines of Arabichandwritten documents. The main goal was to extract each line and distinguish it from others. The main challanges were the overlapping among lines and the non uniform distribution of whitespaces in each document. However, our goals have been achieved by using various techniques organized in three approaches, firstly, we followed the trivial solution to extract lines that are non overlapped using Horizontal Line Projection, af-ter that, we thought of improving the first solutionby using dynamic programming and (Divide andConquer) approach, in which we dealt with eachsub-column alone and obviously, better resultsshown up. Finally, we used Connected ComponentAnalysis to extract lines based on their positionsin the original image, by preparing the image withdifferent preprocessing steps, and the best results were achieved in this part.