Abstract
This paper introduces two robust image watermarking methods using a combination of homomorphic transform and Karhunen Loeve Transform (KLT). The first proposal is denoted as homomorphic-based KLT image watermarking. It works by inserting the watermark during applying the KLT to the reflectance component estimated from employing the homomorphic transform on the host image that will be watermarked. The reflectance component includes significant merits of image, so the embedded watermark with this component is immune to various types of attacks. The second proposal is denoted as homomorphic block-based KLT image watermarking. It works by segmenting the reflectance component of the host image into blocks using spiral scan and adding the watermark to every block during the application of the KLT to each block, separately. The watermark addition using block-by-block principle strengthens the watermark and makes it more immune to attacks. The two proposed methods are compared with homomorphic-based Singular Value Decomposition (SVD) image watermarking method, homomorphic block-based SVD image watermarking method, the traditional SVD watermarking method, and other image watermarking methods. Experimental results ensure that the two proposed methods are immune and secure to different attacks like cropping, scaling, JPEG compression and Gaussian noise without degrading the watermarked image to the same level as the other techniques.