High-Resolution Document Image Reconstruction from Video
Paper ID : 1040-MVIP2020
Hossein Motamednia *1, Mohammad Minouei2, Pooryaa Cheraaqee2, Mohammad Reza Soheili2
2Department of Electrical and Computer Engineering, Kharazmi University
Today, smartphones with high-quality built-in cameras are very common. People prefer to take pictures fromdocuments with smartphones instead of scanning them witha scanner. Due to the limitation of scanners input size, it isdifficult to scan everything with them. Resolution and quality ofsmartphone cameras are not enough to take a picture from largedocuments like posters. In this paper, we proposed a pipeline tomake a high-resolution image of a document from its capturedvideo. We suppose that during the record of the video, the camerawas moved slowly all over the surface of the document from aclose distance. In the proposed method we find the location ofeach frame in the document and we use a sharpness criterion toselect the highest possible quality for each region of the documentamong all available frames. We evaluated our method on theSmartDoc Video dataset and reported the promising results.
Image mosaicking, Image Registration, Image blending, Document image, video
Status : Paper Accepted (Oral Presentation)