A Study on High Embedding Capacity of Image Hiding Based on Run-Length Encoding Scheme

• Main Authors: Shang-Wen Hu, 胡尚文
• Other Authors: Rey-Sern Lin
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/87929218407991712282

碩士 === 玄奘大學 === 資訊科學學系碩士班 === 97 === In this thesis, two modified image data hiding techniques based on Run-Length encoding scheme were proposed. Run-Length encoding approach in image data hiding was firstly proposed by C.C. Chang et al in 2006. Their embedding process was usually associated with three processing steps. The first stage is to convert a binary bitmap image into a series of long bit stream. For secondary stage, an information pair of run count and run value in each run of the same symbol is used to compress the secret information. Therefore, the third stage is referred as embedding process by modifying the least significant bit and adjusting the pixel gray-level in one of two adjacent pixels block. Most binary bitmap images usually have the feature of self-similarity in neighboring region upon which the Run-Length encoding scheme condense the secret image and enlarge the hiding capacity of stego-image. Both in C.C. Chang et al and our first proposed method, a threshold k is required firstly. But, our modification uses an additional bit-information to record whether the run count value is larger than initial threshold k. After that, the hiding capacity of our recipe almost doubles the original one at most and results equal decrease about the PSNR criteria. Due to the mentioned above methods are just do well for binary bit-map image. For random bit stream information, these approaches will be not right for it. An exclusive OR operator, therefore, is applied on the bit stream for clustering together with same symbol in our opinion. This modification can be regarded as a preprocessing before above mentioned works. Since the operator possesses the advantage of invertible, the original bit stream can be re-obtained from transformed bit stream, and it will condense the RLE information pairs. And then, enlarge the embedding capacity of the image data hiding based on the Run-Length encoding scheme. Finally, the whereabouts merit of PSNR and embedding capacity was compared and the experimental results show the efficiency of proposed method for evaluation.