Document capture using a digital camera

This paper covers a method for capturing documents using a digital camera. A typical cheap VGA digital camera (resolution 640 by 480 pixels) does not have adequate resolution to capture text over a page. By displacing the sensor spatially and taking multiple acquisitions we seek to enhance the resolution. There are many approaches in the literature to multi-frame reconstruction. Our approach differs from most previous work in that we make an additional restriction that we wish to enhance the resolution of documents which consist primarily of regions that are spatially piecewise constant (e.g. monochrome text on a monochrome background). We show that this enables us to achieve resolution enhancement that is superior to previously reported schemes. A simulation involving 16 frames of a VGA camera gives excellent rendition of an A4 or legal sized page with 9 pt text