Extraction of Maximum Secret Information Hidden in Approximate Band of Haar Wavelet Transform of an Image

Abstract

Steganography techniques are used to provide security to the sensitive data during transmission and storage by hiding the secret information in the cover media. By applying Haar wavelet transform on the original image the four sub bands (Approximate band (LL), Horizontal band (LH), Vertical band (HL), Diagonal band (HH)) can be obtained. As detailed bands (LH, HL, HH) have lesser information of the original image the more number of LSBs of the detailed bands can be used to embed the secret information. After inserting the secret information in the sub bands inverse haar discrete wavelet transform is applied on the four sub bands. The values obtained after applying Haar IDWT(Inverse Discrete wavelet transform) may have fraction part or negative values. When the fraction part is truncated or negative values are made zero to obtain the pixel values in 0-255 range, the hidden secret information cannot be obtained completely after applying Haar DWT on the received image. In this paper a novel odd-even method is proposed to decrease the number of dissimilar coefficient values of the approximate band and thus to maximize the extraction of secret information hidden in approximate and detailed bands. Total numbers of coefficient values having different values at sender and receiver side are found. Larger the number of dissimilar coefficient values lesser is the secret information extracted. Peak Signal to Noise Ratio (PSNR) value is computed to analyse the quality of the stego image.