Paper Title
Edge Adaptive Image Steganography Based On Variable Embedding

Abstract
Steganography is the art of hiding information in a manner that its very existence is not known. Unlike cryptography which only scrambles the content of the message so that it cannot be read, steganography is a more secure method as it hides the very existence of the message. This paper proposes a new image steganographic technique that embeds secret data only in the edge regions of the cover image while keeping the smoother regions intact. This is done because human visual system (HVS) can tolerate some degree of changes in the edge areas whereas it is sensitive to slight changes in smooth areas. Edges of the cover image are calculated using Canny Edge Detection algorithm. Instead of embedding data in all the detected edges, embedding is applied only in randomly selected edge pixels after doing some pre-processing. For embedding purpose, we use one of the RGB color channels as indicator channel. Secret data bits are embedded to variable number of LSBs of the other two color channels. Maximum intensity color channel is used as indicator (to indicate whether or not data bits are inserted in that pixel). Experimentally it has been found that separating RGB channels into indicator and data channels, causes minimum distortion in the color of a pixel and such changes go undetected by Human Vision System (HVS).