Abstract: Digital forgery and tampering of text documents create an urgent need for content-based document authentication. Observe that main geometric features of characters would approximately stay invariant under small perturbations. A robust document authentication scheme is proposed using a skeleton-based text image hash function. The algorithm is aimed at producing sufficiently randomized outputs which are unpredictable, thereby yielding required properties of image hashing. In the verification stage, the deskewing mechanism base on Hough transform is used to compensate for the distortions induced by rotation or hardcopy operations. Experimental results show that this technique withstands standard benchmark (e.g., Stirmark) attacks, including compression, geometric distortions of scaling and small-angle rotation and common image processing operations. Content-based modifications of text data are also accurately detected.