I am working on a text detection problem from image. The primary task is to identify blocks of text and then use the ones needed. The first step itself is a big mess, because the image are not standard size. Anyways, I have tried a lot of algorithms (Canny Edge Detection, Robert Cross) coupled with Laplacian as well as Gaussian Blur but they are not yielding efficient results. I am attaching some images. I need nutrition and ingredients.