| Abstrakt: |
In today's society, information security and hiding is a key mechanism to protecting one's identity. Image morphing can hide a secret image in a morphed image using a second, reference image. One of the primary methods to make the morphing based information hiding technology practically useful is interactive evolutionary algorithm (IEA). When using IEA, however, the user who is rating the images for their naturalness must rate a large amount of images. During this period of time, it is possible for the user to subconsciously shift the rating criteria, thus skewing the results of the algorithm. To help solve this problem, we propose a k-nearest neighbor (k-NN) reminder, which analyzes the user's previous inputs and creates an estimated fitness rating for images to aid the user in grounding his or her ratings. We ran experiments testing the differences in using the IEA with and without the k-NN reminder, looking to see what impact the algorithm had on the speed of the analysis and the ease of the user's rating ability. Results show that in many instances, the k-NN makes it easier for the user to maintain a standard criterion for the rating of images, with no noticeable impact on the speed of the process. [ABSTRACT FROM PUBLISHER] |
