Seperate Heading, subheading and text from a text document page using open cv

Question

I am building an OCR which could separate underlined Heading, Subheading and text. I am using Tesseract for the same. I have been able to detect the underlined text but I am