A Novel Principal Component Analysis Based Fragile Watermarking Technique for Sensitive Dataset

One of the leading concerns for the cumulative group of content creators, editors, media forensics, military networks and medical communities, is the effective attestation of digital media. With the exponential growth in technology, especially in the field of image processing, mishandling of computerized images has become increasingly customary. Additionally, the tampered images look inconceivably close to the originals. The process is not only quick but also effortless and, in most cases, quick and hassle-free. There exists a provocative need to come up with a streamlined technique to not only confront these difficulties but more importantly, aid in their resolution. Content confirmation, copyright assurance, and insurance are some of the most potent techniques that can help to tackle these perils that exist in the field of computerized data. One such strategy that holds precedence is ‘Computerized picture watermarking’. The watermark information is introduced into a picture. Later on during attestation, the same information is either removed or identified in the watermarked item.The premise of this paper revolves around the use of PCA (Principal component analysis) for watermarking an image dataset. PCA is a machine learning algorithm. The watermarking is done in the projection space, i.e. Eigenvectors. The entire process relies on the effective use of ‘picture pixels’. By controlling the pixel esteems, the watermark can be embedded. In order to maintain the originality of the dataset images, the watermark used is exceedingly fragile. Medicine and forensics serve as the primary market for the deployment of such technique.