Mosaicing of Camera-captured Documents Without Pose Restriction

5 In this paper we present a method for composing document mosaics from camera6 captured images. We decompose the complexity of solving the 8-dof transformation 7 between image pairs into two problems, that is, rectification and registration. This 8 is achievable under a key assumption that sufficient text content forms orthogonal 9 texture flows on the document surface. First, perspective distortion and rotation are 10 removed from images using the texture flow information. Next, the translation and 11 scaling are resolved by a Hough transformation-like voting method. In the image 12 composition part, our contribution is a sharpness based selection process which 13 composes a seamless and blur free mosaic for text content. Experiments show that 14 our approach can produce an accurate, sharp, and high resolution mosaic of a full 15 document page from small image patches captured by a camera with various zooms 16 and poses. 17