A Novel Blind Evaluation Method on Image Scrambling Degree

Abstract

Most of the algorithms for evaluating image scrambling degree depend on the plain image and they cannot achieve blind evaluation. This paper indicates that the ideal scrambled image is similar to the uniform noise in spatial domain, advances a definition of ideal image scrambling which is independent of the plain image and designs a new evaluating index. In frequency domain, the gray relational degree, which is between mean sequence of scrambling image's FFT (Fast Fourier Transform) block and referential sequence, is calculated by gray relation analysis (GRA). Then a blind evaluation approach which combines the approaches of spatial and frequency domains together is presented. Experimental results show that this approach can be operated without depending on the knowledge of plain image and is in good agreement with human vision system (HVS). So this approach can achieve the goal of blind evaluation objectively and possesses high practicality.