A novel semi-blind video watermarking using KAZE-PCA-2D Haar DWT scheme

Abstract

In this research article the digital video watermarking technique is projected through a semi-blind pattern. The proposed method involves frame-spot matching model based on KAZE method at the initial stage, The KAZE method is deployed for matching the edge points of frame-spots with all video frames with the intention to detect the embedding and extract the respective regions. Then the frame entropy blocks are designated and converted by PCA (Principal Component Analysis) blocks. The QIM (Quantization Index Modulation) is employed to quantize the highest coefficient values on each PCA entropy chunks of every sub-band. The single shared secure key is employed to recover the watermarked content. The DWT (Discrete Wavelet Transform) is applied on every single video frame and disintegrate into group of sub-bands. During extraction this is simply reversed; however the KAZE frame-spot is harmonized through each frame edge points. The parameters like rotation, scaling and translation are assessed and the watermarked evidence can be effectively extracted. The proposed pattern is verified using a numerous of video structures and compared with other similar models such as SURF, SIFT, PCA-SIFT, KAZE and perceived high optimal results. The investigational outcomes demonstrated high imperceptibility and high strength against numerous outbreaks like JPEG encoding, addition of Gaussian noise, gamma modification, histogram equality and contrast rectification in both forms of ordinary videos and clinical videos.