A replica consistency detection method for text images based on multi-feature assessment

The replica consistency detection of text images is of great significance in office automation. For example, it can be applied to detect the consistency of the printed documents of commercial contracts and the original electronic documents. In this paper, we propose a method for replica consistency detection of text images based on multi-feature assessment. Firstly, we make a photocopy of the printed document, then apply our method to detect the consistency between the photocopied text images and the original text images. Rather than recognizing each character, we only focus on the measurements of the likeness of these two text images, and no OCR techniques are required. The experiment results indicate that the method has high accuracy. Although the Chinese characters are the main research objects, the experiment results show that the method also can be used for detecting the consistency of Latin letters and Arabic numerals.