Abstract: Although more and more scrambling algorithms are presented，such as Arnold cat transformation，magic square transformation，Hilbert transformation and scrambling based on chaos，they lack a method to evaluate the image scrambling degree.Existing methods aim at a certain scrambling transformation，and the versatility is not better.From uniform distribution，it defines an ideal scrambling model，and according to the character of ideal scrambling，an image scrambling evaluation method is presented.Through comparing the deviation of scrambling degree and ideal state，the scrambling effect can be estimated accurately.Apply Matlab to the simulation experiment of Arnold cat transformation，the result indicates that the method can evaluate image scrambling degree well，and it accordes with people’s vision too.To an image of 256×256，the block of 8×8，16×16 can get the more effect.

摘要： 尽管越来越多的置乱算法被提出，如Arnold cat变换、Hilbert变换、幻方变换和基于混沌置乱变换等，但是针对这些置乱算法的衡量方法却很少，现有的方法都是基于某一种置乱变化而提出的，通用性不强。从图像置乱均匀分布的角度出发，定义了理想置乱状态的模型，并根据理想状态的特点提出了一种图像置乱衡量方法，通过比较置乱效果与理想状态的偏离度来评价置乱效果的好坏。对Arnold cat变换进行了仿真分析，实验结果显示该方法能准确衡量图像的置乱程度，与人类视觉相吻合。并指出对256×256的图像来说，8×8，16×16的分块可以更准确地衡量置乱效果。