Novel approach for image steganography using classification algorithm

Hiding secret data in an any carrier like text file, audio, video or image file is known as steganography [1]. A well defined function is required for steganography for minimizing embedding distortion. This paper proposed a novel technique for image steganography based on DCT and data mining classification techniques. Two 8 bit gray level image of size P X Q and M X N are used as cover image and secret image respectively. DCT is performed on both secret image and cover image. ID3 algorithm is used to find the pixel value or number on where the secret image is going to be embedded. ID3 Algorithm generates the Decision tree to get the proper pixel due to which the embedding distortion will be less. Again two level of security is provided in the system one, before embedding the secrete image on the cover image we apply the Arnold transform, and second key is provided for securing the data. Here key is the pixel number of cover image on which we are going to embed the secrete image.