Abstract: The existence of information to eliminate the problem of inaccurate method to eliminate static exhibits image information based on Arnold transform is proposed. The decomposition of noisy exhibits static images using wavelet transform, to estimate the variance of the static image of each directional subband exhibits the signal through the elliptic directional window; using sub-band energy autocorrelation function to determine the adaptive window to estimate the variance of the static image of each directional subband exhibits the signal, according to inverse wavelet transform for even complex of static exhibits image restoration, the static display image into a plurality of pixel blocks do not overlap each other, static image scrambling exhibits embedded sequence using Arnold transform, threshold division by static exhibits image pixel block level, the biggest change by different levels of pixel blocks image pixel values to determine the maximum range of embedded the secret information.The experimental results show that the proposed method has better visual masking performance, and maintains high image information elimination capacity, and has strong anti attack ability, and the elimination rate is increased significantly.