Digital Image Scrambling Method Of Assessment

Digital image scrambling is a preprocessing method of information hiding, with the purpose of scrambling the pixel position and color. The scrambled image can hardly be recognized, which protects the security of the image information.Currently there exist many image scrambling algorithms, and the performance of those algorithms directly influence the effectiveness of image scrambling and the security of the image transmission. This paper analyzes and simulates some classic image scrambling algorithms, and points out some problems, such as short scrambling period and low security. In addition, some algorithms just take image position transformation into consideration, while others only consider grey level transformation.To solve these problems, this paper designs a dual scrambling algorithm based on two dimensional Fibonacci Transform in the view of pixel position and gray level, which scrambles the pixel position and gray level at the same time. According to the simulation experiment, the image will become disorderly after several scrambling. Compared to the previous methods, our new approach presents a better scrambling effect.How to evaluate the performance of a scrambling algorithm is a hot issue for researchers. Although there have been tens of evaluation methods, no agreed standard has been confirmed. This paper discusses the current scrambling evaluation methods and points out some problems. Some evaluation methods only focus on one aspect of scrambling information, the evaluation of one scrambling method with different images may get different results, and the evaluation performance always has a big volatility and difference.Based on current scrambling evaluation methods, this paper designs another two scrambling evaluation methods. One is easy and fast, with a little calculation amount, based on weighted histogram. The experiment proves the consistency of the evaluation result and the reality, and can reflect the effectiveness of the scrambling times and levels, but it fail to evaluate the scrambling with changed histogram. Another one is based on the average changes of gray level and signal-to-noise ratio, which can evaluate not only the scrambling with changed pixel position, but also the scrambling with changed gray level. The experiment shows that the evaluation results fit the reality.