DEEP LEARNING-BASED WAVELET EMBEDDING FOR COVERT AUDIO OBJECT EMBEDDING IN VIDEO OBJECT STEGANOGRAPHY
Alaknanda S. Patil 1 and Dr. G. Sundari 2
Video watermarking relates to the action of embedding a secret text, audio or file into another video. Generally, the video watermarking is carried out based on two steps, like embedding and extraction phase. Embedding phase embeds the secret information on the video using embedding algorithm, whereas the extraction phase retrieves the hidden data on the embedded video using extraction algorithm. In this research, the video object watermarking is done based on embedding and extraction phase. Here, embedding phase embeds the secret audio message on the object location of video, such that the object location is determined by ‘Shepard convolution neural network’ (‘ShCNN’). In the extraction process, the confidential image is extracted from an immerged video by applying extraction algorithm. Moreover, the embedding of audio signal is hard to process; hence to simplify the process, the audio signal is converted into binary format using Quantum Representation of Digital Audio (QRDA) technique. Furthermore, the training procedure of ‘ShCNN’ is completed utilizing an optimization technique, termed Improved Invasive Honey Badger Optimization (IIHBO) algorithm. Furthermore, an investigational findings provides that the developed plan produced superior results according to the Mean square error Correlation coefficient, and Peak signal to Noise Ratio of 0.028, 0.900 and 46.19, appropriately.