Arabian Journal for Science and Engineering https://doi.org/10.1007/s13369-019-03751-8 RESEARCH ARTICLE - COMPUTER ENGINEERING AND COMPUTER SCIENCE Robust Invisible Digital Image Watermarking Using Hybrid Scheme Dayanand G. Savakar 1 · Anand Ghuli 2,3 Received: 19 July 2018 / Accepted: 2 February 2019 © King Fahd University of Petroleum & Minerals 2019 Abstract Digital image watermarking is a technique to protect copyright of the image owner in the world of digital communication, and robustness is the major property to be addressed effectively. We propose an invisible hybrid watermarking scheme which is composed of blind and non-blind watermarking techniques. First, blind scheme is used as inner watermarking scheme and then non-blind watermarking scheme as outer watermarking scheme. A secret binary image is taken as a watermark and is embedded in an inner cover image using discrete wavelet transformation (DWT) with the help of the blind watermarking scheme in association with predefined binary digit sequence block and gain factor α to get inner watermarked image. Then, this inner watermarked image is embedded into an outer cover image using DWT and singular value decomposition by non-blind watermarking technique to get hybrid watermarked image. On the contrary, to extract the secret binary image, first non-blind watermark extraction and then blind watermark extraction techniques are used. From the experimental results, it is shown that this hybrid watermarking approach is robust against—rotation, JPEG compression, salt and pepper noise, Gaussian noise, speckle noise and Poisson noise. Keywords Secret binary image · Hybrid watermarking · Inner cover image · Outer cover image · Inner watermarked image · Blind · Type-II non-blind 1 Introduction In the field of digital watermarking, many researchers are trying to reduce the gap between robustness, fidelity and capacity requirements. However, more emphasis is given for devising robust digital image watermarking scheme for copy- right protection. The image with embedded watermark is called watermarked image. This watermarked image could be published, and the ownership of a suspected image can be claimed by retrieving the watermark from the water- marked image. A robust digital watermarking scheme should resist destruction from standard image processing and mali- B Anand Ghuli anandghuli@gmail.com Dayanand G. Savakar dgsavakar@gmail.com 1 Department of Computer Science, P.G. Center, Rani Channamma University, Vijaypur, Karnataka 586108, India 2 Department of Computer Applications, B.L.D.E.A’s V.P.Dr.P.G.Halakatti College of Engineering and Technology, Vijayapur, Karnataka 586103, India 3 Visvesvaraya Technological University, Belagavi, Karnataka, India cious attacks. Suppose the watermark (WM) of size p × q is to be sent to the receiver. This watermark is embedded into the cover image (CI) of size M × N . This embed- ding process yields digital watermarked image (DWI), i.e., WM + CI = DWI of size M × N . During the transmis- sion through communication channel, this DWI is subjected to numerous types of attacks attempting to extract the trace of WM from DWI. Different set of frequency domain trans- formations were used by many watermark embedding and retrieval techniques to attain robustness. The digital water- marking process should be very much transparent, and this transparency can be measured by computing the level of distortion by suitable measures between CI and DWI. The technique and methodology of embedding WM into CI should be such that it should protect the WM from various types of attacks (i.e., robust) [33]. The digital watermarking schemes may be divided into two, namely blind scheme and non-blind scheme [33]. If the sender of a message is embed- ding the WM into CI using only suitable key and the same key is used during WM retrieval from DWI is known as a blind scheme. If the copy of original cover image (*note: This con- cept is used in our non-blind technique and the term copy of inner cover image is used) is needed at the receiving side to 123