Secret Speech Hiding in Image Based on Quantization Level Modification

Abstract

Data hiding is a form of steganography method that hides secret data in a cover medium like an image, audio or video. In this paper, a method of hiding speech data in a colored image is presented. The main method used in this proposal depends on the technique of quantization level modification with Discrete Wavelet Transform (DWT) to process the cover medium. The latter is used to extract from the RGB of colored images its three bands namely; Low-High (LH), High-Low (HL), and High-High (HH), to be used for embedding the data. As well as Discrete Cosine Transform (DCT) compression is used to compress the speech data by 40% ratio. The embedding method mainly depends upon Absolute Moment Block Truncation Coding (AMBTC) technique for manipulating the binary format of speech. In order to increase the complexity of this proposal, the cover image data is scrambled by using a permutation key that is generated by the Henon map. This proposal also can be utilized to hide any password needed to be extracted by the end-user. The experimental result shows that the average of mean square error (MSE) is 12.16186, peak signal to noise ratio (PSNR) is 33.34402, while the structural similarity (SSIM) is 0.97852.