Alam Naved, Yadav Rajesh
Abstract: In this era any type of digital media such as image, text, audio and video, all are easily accessible and transferable through the use of high speed internet. As the use of internet increased, the need of security and authenticity also increased. To secure the multimedia data such as image, audio, text and video researcher has developed a watermarking technique which gives a watermark embedding and watermark extraction algorithm, and later it is used for proof of ownership. Here we are proposing a technique on video watermarking using 2D DWT and 2-level SVD technique. In this paper first we are taking a video which is decomposed into number of frames and embedding a watermark image on each frame. First 2-D DWT is applied on each frame. Dwt decompose each frame into low frequency, mid frequency and in high frequency (LL, LH, HL, HH) band then we applied SVD on LL and HL sub-band called it dual band. SVD convert it into three matrices as U1S1V1 of single matrix. A watermark image is taken which converted in to gray scale from RGB scale then resized it. And embed this watermark image into host frames with some scaling factor. After that we again applied SVD on this watermarked frames which further convert this single matrix into three matrices as U2S2V2 now multiply S2 matrix with U1 and V1 matrix component to make it more secure. To demonstrate the authenticity of this watermarked video we applied some attacks such as Gaussian filtering, median filtering, frame rotation, contrast adjustment and sharpness attack which show its PSNR and NCC value in comparison with the original video.
Keywords: DWT, SVD, Matrix, Attacks, PSNR