I am planning to write an OCR for english handwritten, I did review on text line extraction but I could not figure out what is the reliable algorithm to extract handwritten text lines. My documents are not printed so my text lines can have over lap or skew also I can have a lot of noses. These are the algorithm I find used commonly: extract the projection and find the baselines or either use the connected components or clustering algorithm. Some researcher also have used Level set or snakes algorithm to extract lines from handwritten documents. I really don't know which method should I use or from where I should start. At this stage I am assuming that my documents are skew and rotation free but the text line may have skewed because they are handwritten and I don't need to be worry about noise or any other pre-processing.
One more question, does Opencv has any function to work with text in order to make the lines or words extraction easier?