A high-capacity text watermarking method based on geometric micro-distortion

Motivated by the increasing necessity of copyright protection for textual documents, we propose an automatic generation method for high-capacity text watermarking that leverages Chinese character structure and variable font features. The framework was developed by hiding a watermark in Chinese textual files. By replacing the original characters in the documents with ideologically equivalent but morphologically different glyphs generated using our proposed method, we embedded the watermark into documents. Previous watermarking methods mainly face the challenge of low embedding capacity and the replacement glyph was not automatically generated. To address this problem, we focused on a new glyph variation-generating approach. we first devised four types of geometric micro-distortion methods for Chinese characters, including changing the font-weight, panning the stroke components, adjusting the thickness of strokes and altering the spatial layout of components. Glyph variations were then generated by applying one or more than one of the four types of micro-distortion methods. Using variable font features, numerous glyph variations can be automatically generated in a feature-continuous manner, which provides our algorithm with a relatively high capacity. The experimental results indicate that our approach has better capacity and imperceptibility performance than those of state-of-the-art Chinese text watermarking approaches.