Hybrid SOFM–MLP Neural Network for Steganalysis to Detect Stego-Contents in Corporate Emails

Abstract

A hybrid approach of cryptography, data compression and steganography has been proposed in this paper. Motivation behind this research is to provide a smart image steganographic technique which must be capable enough to provide better quality stego-image with a high data hiding capability. Proposed approach is a LSB based approach in the field of image steganography. Maximum data hiding capability of proposed approach will be evaluated from kekre's algorithm. Proposed approach hides data in the upper LSB bit only when its adjacent LSB bit of all the pixel have conceived a bit of secret data for better quality of the stego-image. LZW compression scheme is used to optimize the size of secret data, it will enable a personal to hide approx 2 times more data in a cover-image. This approach is secure against the detection attack and its stego-image is totally indistinguishable from the original image (cover-image) by the human eye.