KEY Genaration Forimage Scrambling Using Voiceprint

Abstract

This paper presents a new algorithm to scramble color image using voiceprint and linear predicative coding (LPC). The speech signal pass through pre-processing stage which includes sampling and segmentation into many frames. All frames are windowed using rectangular window and fed to linear predicative predicator, the linear predicator is used to obtain the coefficient of the pth order all-pole vocal tract and it predicts the current sample of the speech signal from linear combination of past samples. Levison Durbin (L-D) procedure is used for each speech frame to find Lp coefficients, reflection coefficients and predictor error. For scrambling color image, key will be generated manually; by using the LPC coefficient, by ascending all the LPC coefficients and compare each coefficient with all pixels of the color image. When LPC coefficient is similar to the pixel, the pixel will be replaced by that coefficient. So that pixel will be send in random sequence and the color image will be scrambled by using voiceprint (LPC) coefficients. Descrambling will be done in reverse procedure. Scrambling process is simulated using MATLAB version 7.06.324(R2008a). Many tests are done with different speech signals and color image, SNR, correlation will founded good results.