A Novel Image Steganography Method Using Fuzzy Logicand Edge Detection

Abstract

Steganography is one of the information hiding techniques that hides a message inside another message without drawing any suspicion. For hiding messages, various types of media are used. Image steganography is the technique that uses an image file to conceal information. In recent years, different methods have been proposed, which combined steganography and edge detection, have been proposed. This thesis presents a novel Image Steganography method using least significant bit (LSB) and fuzzy logic. Firstly, gradient type-1 fuzzy logic (T1FLS) edge detector has been proposed to make disclosing the existence of a secret message a hard operation. The proposed system processes the image in two phases (named fuzzy phase and embed phase). In the fuzzy phase based on the gradient approach and T1FLS, the edge detector is calculated. In the embedding phase, exploiting the edge image that has been obtained from the previous phase in embedding more secret bits in edge pixels than in non-edge pixels. The proposed system is developed on two sides, the sender’s side which deals with the embedding process, and the receiver’s side which deals with extraction processes. Secondly, the gradient T1FLS edge detector has been improved by using gradient type-2 fuzzy logic(T2FLS) edge detector due to their ability to handle the high level of uncertainty present in images. The enhanced system processes the image in two phases (called the fuzzy phase and embed phase). The enhanced gradient T2FLS edge detector has the same steps as the gradient T1FLS edge detector, except for the use of T2FLS instead of T1FLS in the fuzzy phase.Many experiments were conducted to measure the performance of the proposed methods. For the proposed gradient T1FLS edge detector, the experimental results demonstrate the performance of the proposed T1FLS on six 128×192 RGB color images from the BSD300 dataset. For the enhanced gradient T2FLS edge detector, three experiments were conducted on different image datasets based on the image size. In the first experiment, the proposed T2FLS will apply to six 128×192 RGB color images from the BSD300 image. In the second experiment, the proposed T2FLS will be implemented on six 256×256 RGB color images from the USC-SIPI image dataset. And in the third experiment, the proposed T2FLS will be implemented on six 512×512 RGB color images from the CSIQ image dataset. The PSNR and HVS have been used to measure the quality of the stego image in each experiment. When the results of the proposed method were compared with previous studies, the results showed that the proposed system provides better stego image quality, as well as higher embedding capacity than previous works. Metrics like peak signal to noise ratio (PSNR) and the human visual system (HVS), have been used to measure the quality of the stego image.