A Hybrid LWT-DWT Digital Image Watermarking Scheme using LSVR and QR-factorization

The digital image watermarking is an art of hiding secret image into source image. The robust approach of this watermarking practice is to trade-off image quality (invisibility) and embedding capacity. In this proposed article we present an efficient and hybrid approach that integrates features of lifted wavelet transform (LWT) and discrete wavelet transform (DWT) based on linear support vector regression (LSVR) and QR-factorization for watermarking. Initially, the QR-barcode (watermark image) grayed to achieve maximum watermarking capacity. The correlation feature of LWT is used to decompose image into nonoverlapping blocks. The QR factorization is further decompose into Q and R matrices. The LSVR is obtain input samples from R-matrix for training and learning purpose. In the second segment, we implement the DWT based multichannel fake-proof authenticity mechanism. Precisely the integrated hybrid approach produces less distortion rate. The experimental results is analyzed with other models and offers high reliability on watermark embedding and authenticity along with less computational cost.