IMPROVED ROBUST IMAGE WATERMARKING WITH POLAR HARMONIC TRANSFORM AGAINST TRANSLATION ATTACK

• Main Author: Hanifan, Mohamad
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/50780
• Institution: Institut Teknologi Bandung

Robust image watermarking (RIW) is a type of image processing that is used to embed information / watermarks into digital images to prevent piracy. The insertion process must not cause the image to change and the embedded watermark must not be damaged even though the image is attacked. Some of the early methods used to perform RIW include DCT, DWT, or SVD which are very good at overcoming types of signal processing attacks. Meanwhile, to overcome geometric attacks such as rotation and scaling, radial-based transformations such as ZMs / pZMs, RHFMs, or PHT can be used. Among these transformations, the Polar Harmonic Transform (PHT) has the advantage of being less complex and thus simplifying the computational process. The disadvantage of RIW with PHT is that it cannot overcome this type of translation editing because it causes the image center point to shift, even though this point is needed as a reference point in carrying out transformations. In this study, this type of translation editing will be overcome by using the ORB feature matching to calculate the position of the new center point after the image has been translated. The test results show that without the proposed method, a translation of 3 pixels can cause watermark damage as much as 30%. This proposed method can calculate the center point of the transformation accurately as long as the portion of the image lost due to translation is less than 43%. In the translation range of 12% in the vertical and horizontal directions, with this proposed method the number of damaged information bits drops from about 30% to 10% for the translation type which causes part of the image to be lost, and 0% for the translation type without causing part of the image to become lost.