| dc.description.abstract | The advance of information technology such as marketplaces, e-money, and
insurance are required to upload an identity card file for account verification or for
electronic payments. The dissemination of this personal information could be
misused by a third party who may able to break into the account users and steal the
data. Therefore, protection of this personal data is necessary for authentication. In
this study, the proposed technique for protection is watermarking. This
watermarking is in the form of a binary image which will be embedded in the
identity card file using Discrete Cosine Transform method. Discrete Cosine
Transform is a method to change the image from the spatial domain to the frequency
domain. Image transformation to frequency domain produces a DCT coefficient
value that consists of subband frequency. To meet the criteria for good
watermarking, embedding analysis is carried out in the frequency subband (low,
middle, high, low-middle, and middle-high for each color component. Pixel
coordinates the embedding watermark that is different for each 8×8 DCT block for
there is an order of values selected for the DCT coefficient. Imperceptibility and
robustness analysis were conducted on 10 scanned images with a resolution of 100
dpi and 200 dpi in BMP format. The results showed that the PSNR value is ≥ 52
dB with a (watermark strength) 0,1 and the average SSIM score ≥ 0,9. In order to
maintain a balance between imperceptibility and robustness, a good watermarking
should be conducted at a high frequency on green component while for a red
component is at low-middle frequency those are susceptible of attacks in the form
of brightness +20 and contrast +50 with an average of NCC value ≤ 0,85. | en_US |
