Paper Form Image Recognition Based on SSIM
Chen Zhong, Tianxiang Zhou, Ke Wang · 2019
This paper introduces a method of filling information recognition for scanned image of paper form. The cross point coordinates are extracted by Strip-like corrosion and expansion of the binary images. The form image base frame is constructed using a rectangular fit based on the Hough transform. Using the image pan, rotate, and scale algorithms, the template form image base frame to the scanned form image base frame can be matched. The filling grid area is determined based on the structural similarity algorithm (SSIM) and the filling information is thus obtained. The experimental results show that this method can accurately, quickly and efficiently read the batch paper form filling information digitally, and has strong robustness.