Steganalysis ppt
- 1. Steganalysis By Om Shivam Btech CSE+LLB(Hons) Cyber LAW 4th year University of petroleum and energy studies
- 2. Image Steganalysis Algorithm are two types:- (1) Specific For the specific type of stegnographic algorithm. (2) Generic This is the valid for all type of stegnographic algorithm.
- 3. Specific Image Steganalysis Algorithm Image steganography algorithms are more often based on embedding mechanism called least significant bit (LSB) embedding Each pixel is represented as 24-Bitmap value composed of 3 byte representing the R,G and B values for the three primary colors red, green and blue respectively. Higher RGB values for pixel implies lager intensity for instance , a pixel p represented as FF FF FF16 as it is composed of all these primary color so this is white. In we make use of LSB as there is minor change in pixels which is not perceptible of human eye like FE FE FE16 would make color darker by the 1/256 factor.
- 4. Several Formats (1) GIF (Graphic interface formats) (2) BMP( Bitmap) (3) JPEG(Joint photographic Exchange Group) Each formate behaves differently when a message is embedded in it.
- 5. Palette Image Steganalysis It is for the primary used for the GIF images Gif property It support 8 bit pixel of the color of the pixel referenced from the palette table of up to 256 distinct colors mapped to the 24 bit RGB color space. What are changes happens in Steganography? • Message embedding brings great change in the palette color of the pixels.
- 6. How Steganalysis applies to GIF? The Steganalysis of a gif stego image is conducted by performing a statistical analysis of the palette table vis-à-vis the image and the detection is made when when there is an appreciable change in entropy ( a measure of the variation in the in the palette colors) the change in entropy is maximal when the embedded message is of maximum length.
- 7. Raw image Steganalysis It is basically used for BMP images that are characterized by a loseless LSB planes. Here LSB embedding on image causes flipping of the grey scale values. The embedding of the hidden message is going to result in the average of frequency occurrence of the pixel with two Grey scale values. For ex:- One image has 20 pixel with one grey scale values and other 40 pixel with the other grey scale value. The pixel around to be 30 pixel.
- 8. Steganalysis of BMP Dumitrescu proposed the algorithm, algorithm assumes that an image to be made up of horizontally adjacent pixels and classifies the set of all such pixel pairs(a,b) into four subsets depending on whether a and b are odd and even. whether a<b, a>b , or a=b What happens when the BMP image is embedded? The pixel values get modified when the message embedding is done in the LSB planes leading to the membership modification across the four subsets. The statistical analyses on the change in the membership of the pixels in the stego image leads to the detection of the length of the hidden message.
- 9. For JPEG formats Two well known steganography algorithm for hidden secret messages in jpeg images are 1. F5 algorithm 2. Outguess algorithm
- 10. F5 Algorithm It uses matrix embedding to embed bits in the DCT( Discrete Cosine Transform) coefficients in order to minimize the number of changes to a message. F5 mutates the histogram of DCT coefficients. Frendrich proposes a technique for unaltered histogram to find the number of changes and length of secret message. The process involves cropping the images into four column and then applying quantization table to recompress the image. The resulting DCT will image close ot the original one. Frendrich also proposed for the outguess algorithm What outguess algorithm do It is basically in random walk embed the message bits in the LSB of some of the DCT coefficients The other DCT Coefficients are then adjusted to keep the original histogram intact.
- 11. Here the nature of the discontinues in the DCT coefficients is widely employed to gauge the size of hidden message.
- 12. Generic image Steganalysis Algorithm Here we focus on the features that are monotonic and changes statistically as a result of message embedding. Avivas use a set of image quality metrics(IQM) to develop a discriminator algorithm that differentiates cover image from stego images. Here it is rather for analysing the image quality and performance of the algorithm it is used for the Steganalysis. The ANOVA( Analysis of the variance is used to test to rank the IQM based on their F scores and identify the embedding message. The main motive of employing this method is to detect the distortion at different level of sensitivity
- 13. For example The means square values for the human visual system (HVS) weighted errors demonstrate more sensitivity to pure blur While the gradient measure responds to change in the texture and image periphery. Avicas proposes another approach that analyses the seventh and eight bit of the planes of an image measure their binary similarity
- 14. Audio steganography algorithm 1. Low bit encoding 2. Phase coding 3. Spread spectrum coding 4. Echo data hiding
- 15. Low bit encoding Here the secret message is embedded on the least significant bits of each sample. Phase coding It is based on the facts that the phase component of sounds are not as perceptible to the human ears as noise is This lead to the inaudible encoding in terms of signal to perceived noise ratio and the secret message is send in the audio signal.
- 16. Spread Spectrum coding Ss randomly spread the bits of secret data message across the frequency spectrum of the audio signal. Echo hiding Echo into the discrete audio signal Here for successful hiding three parameters of the echo need to be altered Amplitude, decay rate and delay time(off set) from the original signal
- 17. Audio Steganalysis algorithm Zeng proposed algorithm to detect the phase coding steganography based on the analysis of phase discontinues and to detect echo stegnogarph based on the statistical moments of peak frequency A statistical analysis of the phase difference in each audio segment can be used to monitor the change
- 18. Video Steganalysis This technique applies to frame to frame budia created framework based on Steganalysis of Gaussian spread spectrum These are two essential blocks (1) Water marking attack stage to eliminate the cover media from possibly watermarked stego media. (2) A pattern recognition stage for the detection of stegnographic activity.