In the recent years, there has been a great deal of interest in developing a secure algorithm for hiding information in images, or steganography. There has also been a lot of research in steganalysis of images, which deals with the detection of hidden information in supposedly natural images. The first section of this thesis reviews the steganography algorithms and steganalysis techniques developed in the last few years. It discusses the breadth of steganographic algorithms and steganalytic techniques, starting with the earliest, based on LSB flipping of the DCT coefficients, to more recent and sophisticated algorithms for data hiding and equally clever steganalytic techniques.

The next section focuses on the steganographic algorithm, MHPDM which was first developed by Eggers and then modified by Tzschoppe, Bauml, Huber and Kaup. The MHPDM algorithm preserves the histogram of the stego image and is thus perfectly secure in terms of Cachin's security definition. The MHPDM algorithm is explained in detail and implemented in MATLAB. It is then tested on numerous images and steganalysed using Dr. Fridrich's recent feature-based steganalytic technique. The thesis concludes with observations about the detectibility of MHPDM using feature-based steganalysis for different payloads (embedded message lengths).