A Scrambled Image Blind Evaluation Method Based on Space and Frequency Domain

Abstract

Most of the image scrambling degree evaluation algorithms rely on the statistical features of the original image, which cannot achieve blind evaluation. Based on the uniform distribution model of the statistical characteristics of ideal scrambled images, a blind evaluation algorithm of image scrambling degree combining space and frequency domain is proposed in this paper. In the space domain, the uniform distribution characteristics of the gray histogram of the ideal scrambled image are used, and the uniform distribution characteristics of the Discrete Fourier Transform (DFT) spectrogram in the frequency domain are combined with the gray correlation analysis theory. The two are weighted to realize the space and frequency domain evaluation of the scrambled image performance. The experimental results indicate that the evaluation algorithm in this paper can consider the performance evaluation of both pixel value and pixel position dislocation, avoiding the disadvantage of ineffective spatial domain evaluation when only the pixel position scrambling is performed. It has very sensitive to the histogram distribution and frequency domain features of encrypted images, and has good agreement with Human Visual System (HVS). The original image is completely independent, and it enables blind evaluation objectively.