M.A. Thesis:

Investigation and improvement of LSB and DCM Steganography algorithms in digital images and classification using MLP neural network

Abstract: Due to the increasing speed of the new processors and the use of these processors on the one hand, and on the other hand, significant advances in new algorithms such as parallel processing and cloud computing, as well as the development of artificial intelligence algorithms, such as neural networks, information steganography process, with reduced probability of simple diagnosis will not be. Also, increasing the bandwidth and the speed of sending information in new applications. The ability to send more volume information is easily provided in new technologies. In the image hiding, a signal hidden in a spatial domain or one of the frequency domains, such as the discrete cosine transform, Fourier, and wavelet, can be hidden. Concealment techniques in the area of conversion show a greater resistance to various attacks than those in the area of the field, because when the image is converted inversely, the secret image is unbiased. It will be distributed throughout the image, so it will be very difficult to read and modify it for the attackers Several cryptographic algorithms are provided for different image structures. In general, image cryptographic methods are embedded in bitmap algorithms and extraction algorithms. Some of the common ways in steganography image files are: Substitution of the least value bit (LSB), threshold correlation, correlation based on comparison, wide spectrum method Comparing Medium Bandwidth, Wide Range in the Dynamic Duct Waveform Domain, In light of past work done in this field, we intend to explore and develop LSB and DCM steganography algorithms in digital images. For this purpose, we will use cryptographic methods that include discrete wavelet transformations, and we will use the MLP neural network to classify these images.

Keywords: