DCT VERSUS WHT FOR FRAGILE IMAGE WATERMARKING
U Anna · 2004
In the signal transmission process one has to take into account also potential errors that can appear. So on the receiver’s side we want to be able to detect these errors. Digital watermarking seems to be a possible solution of this problem [1]. It means that on the transmitter’s side we add some redundant information a fragile watermark. The receiver tries to extract this watermark. Since we embed a fragile watermark, an error in the data stock causes watermark’s destruction and this fact uncovers the data transmission error. The algorithm described in the rest of this paper was originally proposed for video applications [2] and the watermark was embedded into the quantized DCT coefficients. We test not only DCT spectral coefficients but also WHT spectral coefficients for fragile watermarking of still images. In the analysis we concern with two types of errors: (1) The pixel’s value was set to zero (it may represent the loss of the pixels) and (2) Removing the rows or columns from the image. The loss of the column or of the row corrupts synchronization between the transmitter and the receiver. So the output of the detector should be wrong blocks after deleting a row or column. On the receiver a demand should appear for retransmission of these blocks of the image. Also in this case, like for every kind of the watermarking algorithm, we have to make a compromise between the probability of errors in the communication channel and visual quality of the watermarked image. The following part includes a description of the principles of embedding the fragile watermark and of the detection of errors. Finally, we show some experimental results.