A DWT-based Watermarking Algorithm Robust to Geometric Attacks

TL;DR: The algorithm uses the characteristic vector with a HASH function to embed or extract the watermarking and is robust to rotation, scaling,transla- tion, and bending of images.

Abstract: Most watermarking algorithms are not robust to geometric attacks.This paper presents an algorithm based on the discrete wavelet transform(DWT)to embed and extract watermarks.A watermarking is embedded in four stages.First,the original image is decomposed using DWT.Secondly,the watermarking is embedded into the approaching sub-band image in DWT domain.Thirdly,a characteristic vector of the image is acquired using a dis- crete cosine transform(DCT).Finally,a binary logic sequence,or secret key,is produced using the characteristic vector and the watermarking.The watermarking can then be extracted in three stages.First,the test image is decom- posed using DWT.Secondly,a characteristic vector is obtained in the DCT domain.Finally,the watermarking is ex- tracted using the secret key and the characteristic vector.The algorithm uses the characteristic vector with a HASH function to embed or extract the watermarking.Results show that the algorithm is robust to rotation,scaling,transla- tion,and bending of images.Robustness against rotation is demonstrated up to 40°and against scaling down to 20% of the image's original size.Furthermore,robustness against whirly bend is demonstrated for angles up to 100°.