Extrinsic Signature Embedding in Text Document Using Exposure Modulation for Information Hiding and Secure Printing in Electrophotography

Printer identification based on a printed document can provide forensic information to protect copyright and verify authenticity. In addition to intrinsic features (intrinsic signatures) of the printer, modulating the printing process to embed specific signature (extrinsic signatures) will further extend the encoding capacity. Some features generated by modulating EP process such like raggedness of the edge in the text, can be designed and utilized to help develop the extrinsic signature. By increasing the level of modulation, which increase reliability of detection and thus capacity, we can drive the extrinsic signature to a point without perceptual degradation of the image quality. In this paper, we will investigate embedding extrinsic signature in text documents using laser intensity as a signal modulation source and develop extrinsic signatures using both frequency as well as amplitude modulation. Preliminary experimental results showed that it is feasible to embed detectable extrinsic signatures in text characters without degrading perceptual text quality. Introduction Banding is one of the image artifacts for electrophotographic (EP) printers. Due to its origin within the EP process, it can also be viewed as an intrinsic signature of the specific printer. Modulating the EP process to generate banding signals that are below the human visual threshold but can be detected by effective detection approach can further extend the signature capacity. In our previous work, we have reported techniques that modulate the exposure of the EP process through laser intensity modulation and exploit the reduced contrast sensitivity of the human visual system at higher spatial frequencies. Unperceivable banding signatures of different frequencies and amplitudes can be embedded into an image and reliably detected. Given the amount of text documents involved in forensic analysis, it is desirable to expand the ability to embed robust extrinsic signatures in text documents with high technical barrier of entry. The absence of large mid-tone areas makes it difficult to capture suitable signals for banding detection for text documents. Other text features such like raggedness can to be utilized to develop other types of extrinsic signatures. In this paper, we will show that by modulating the laser exposure of the EP process and exploiting the features of text, extrinsic signature of different frequencies and amplitudes can be embedded into text and reliably detected without perceptual degradation of printing quality. A HP color laser jet 4500 is used as the experimental platform for this study. In the following sections of the paper, the method for embedding extrinsic signatures into text documents without degrading the perceptual image quality is discussed and corresponding experimental results are also provided. The summary and conclusion are given in the last section. Method for Embedding Extrinsic Signature Unlike halftoned images, majority of the texts are composed of individual characters with white space separating lines of texts. If laser exposure modulation can be extended to each line of characters, as shown in Fig. 1, and reliably detected, it will significantly improve the encoding capacity. As with embedding extrinsic signature in halftoned images, minimal visual impact and robust detection are the key issues for embedding extrinsic signatures in text documents. Figure 1. Embedding extrinsic signature in text A. The Effects of Modulating Laser Intensity Figure 2(a) shows the effect of modulating laser intensity in different types of images. The first line is printed without any modulation while the second and third lines are modulated with banding frequencies of 20 cycles per inch and 40 cycles per inch, respectively. If the modulation intensity is strong, the banding will be significant in halftoned grey scaled images shown in the right hand side of Figure 2(a). However, for characters, because of the absence of mid-tone area, the ability to detect the embedded signals is limited. However, the negative visual impact of ragged edges on text characters due to high modulation intensity is evident. Figure 2(b) shows the magnified letter I in the third line of Figure 2(a). Similar to embedding extrinsic signatures in halftoned images, the edge raggedness is a possible signature but the trade off between signature detection and image quality is the main issue that needs to be factored into the embedding strategy.