Semi-Fragile Auto Image Inpainting based on Structural Similarity

• Main Authors: TSAI CHIEN AN, 蔡建安
• Other Authors: 洪國銘
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/02914975517298358782

碩士 === 開南大學 === 資訊學院碩士在職專班 === 102 === Due to the construction of a comprehensive network, digital cameras and smart phones in general, the transmission of digital images sharp increases. However, the problem is easy to loss data when digital image transmission. And distortion caused by compressed transmission is also need to be considered. Therefore, this paper proposes a Semi-Fragile Auto Image Inpainting based on Structural Similarity. It is three stages including self-embedding, certification, repairing. It makes a stego media and self-embeds for original picture at stage of self-embedding into medium frequency of discrete cosine transform. In certification stage, to extract original digital watermarking and contrasting image. And to certificate in structural similarity index with original digital watermarking. Doing repair stage if damaged or modified. Using the extracted digital watermark of suspicious images in the repair stage, and look-up table make it inverse halftoning. Then use bilinear interpolation to repair damaged and modified region. The proposed method is to use a random number seed break up the watermark, It can spread blocks of attack to the whole image to reduce damage of perception to the human eye, and DCT convert watermarks to embedded in the frequency domain, so they can more effectively resist the destruction of JPEG compression technology and the use of information hiding technique for halftone images embedded in the original image at the time to repair the destruction of the image, experimental results show that the proposed method can effectively resist the reasonable compression, the image have some repair capability when it is damaged.