Cognitive Based Data Classification in Image Steganography for Secured Data Transmission

Steganography is the art of hiding information within cover medium in such a way that it is undetectable. The main problem with steganography in the current system is that it does not differentiate between a highly sensitive document and non-sensitive document. Such a kind of steganography system fails when image manipulations are performed. A novel method to perform steganography based on the sensitivity of the data is developed to overcome the above stated problem. The sensitivity of the data is decided using the K-Nearest Neighbours (kNN) algorithm. As data classification is one of the most essential tools needed for data security such a sensitivity-based labelling is required before hiding data in a cover medium. One of the most important features of data classification is to find duplicate data to cut storage and backup costs. The decision to select the right algorithm is based on the security level assigned to data, size of data and the Peak Signal to Noise Ratio (PSNR) of the algorithms. With the kNN algorithm, an average accuracy of 90% has been achieved for data classification. In the data hiding aspect of the project, PSNR values ranging from 48dB to 70dB have been obtained. Data classification results in accurate results and the steganography that follows presents clear images without distortions.