Multimedia Tools and Applications https://doi.org/10.1007/s11042-023-15349-2 Robust digital color image watermarking based on compressive sensing and DWT S. Prasanth Vaidya 1 · P. V. S. S. R. Chandra Mouli 2 Received: 15 July 2022 / Revised: 6 April 2023 / Accepted: 15 April 2023 © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2023 Abstract In this paper, a powerful digital watermarking algorithm derived from compressive sensing and discrete wavelet transform to provide authentication and preserve copyright protection for color images. Initially, host color image is converted into YCbCr. Then Y “luminance image” is decomposed with wavelet transform and its low-frequency subband is selected. The watermark image is divided into non-overlapping blocks which are compressed. Finally, the compressed watermark blocks are embedded in Y. In extracting the watermark, L1 opti- mization algorithm is utilized. Investigated results convey that the developed algorithm has better performance, high concealment and high durability over various attacks of image. Keywords Digital watermarking · DWT · Compressive sensing 1 Introduction In recent times, network and information technology play an important role in data exploita- tion, national defence security, medicine and other fields. Information safeguarding is a prominent role in securing the multimedia from intruders [12, 26, 27]. To provide authentic- ity to owners of the data and to protection multimedia copyright data, digital watermarking is the best solution [8, 23, 28]. Recently, an encryption algorithm using compressive sensing is proposed [5, 6]. The compressive algorithms are given by Baraniuk and Bran- nock [2, 3]. Watermarking schemes based on Compressive Sensing (CS) are used in tamper detection [11, 34] and also helps to improve watermark robustness with a reduced set of measurements. Through a random measurement process, the signal can be exactly recov- ered from incomplete data using CS. This method can efficiently resist geometric attacks through measurement matrix. As measurement matrix considers important and unimportant data equally, even with a loss of few data, original data can be reconstructed perfectly. High  S. Prasanth Vaidya vaidya269@gmail.com P. V. S. S. R. Chandra Mouli chandramouli@cutn.ac.in 1 Aditya Engineering College, Surampalem, India 2 Department of Computer Science, Central University of Tamil Nadu, Thiruvarur, Tamil Nadu, India