Abstract:
Multimedia Information Relay and Garbled Embedding, MIRGAE, describes watermarking for images and audio files. Watermarking, the art of secretly embedding data, has become exceedingly popular during the past few years.
This project explains the techniques used for image and audio watermarking. The watermarks are coded, embedded and recovered from the host signal. Typically the containers chosen are images, audio and video, but this project emphasizes on images and audio as the host signals. The watermarked images are added with noise, compressed and resized to simulate attempts to remove the watermark and also to figure out the most robust technique.
Domains such as Spatial, Frequency and Wavelet are described in detail. Implementation of techniques for each domain is carried out separately and the results of each, based on certain metrics, are discussed.
A Discrete Cosine Transform based technique for audio watermarking is used. The communicated audio signal is hidden in White Gaussian Noise in the DCT domain; methods used in transmitter and receiver side are discussed in detail. A Band pass filtering method is also discussed in detail. In which the host and message undergo a series of filtering including low pass, high pass and band pass filtering; methods used in transmitter and receiver side are discussed in detail.
