A novel automatic deforming for smartphone image scanning using gradient-based corner detection

The paper proposes an automatic algorithm to deform document images resulting from document scanning using smartphone camera. The algorithm integrates computing image gradient, edge detection and geometric transformations. At first, the gradient operators are implemented to estimate the location of edge pixels. Then the scanned document corners are found based on the edge map. Finally, geometric transformations combining projective mapping and bilinear interpolation are used to deform the image, which makes the image well-balanced and easier to read. The algorithm demands simple computation, which helps to obtain low complexity and short running time for smartphone application. The effectiveness of the proposed algorithm for deforming image is demonstrated and the evaluation results are reported in this paper.