Neural network based CT-Canny edge detector considering watermarking framework

In this paper, the application of the neural network based watermarking framework is considered in the area of transformations. The logo information is embedded in the edge of the Contourlet transform. The canny edge detection is applied to detect the edge associated with the transform coefficients. The genetic algorithm has been used to choose the transform level and watermarking intensity. The genetic algorithm selects the best transform level and watermarking intensity based on the lowest error in extracting logo information for different default attacks. Of course, it should be noted that depending on the capacity of the logo, the number of subbands will be selected. In this paper, two methods of the differential and neural network are used to extract the logo and then the two methods of extraction are compared in terms of error extraction. The approaches of the embedding and the de-embedding in case of learning algorithm of the neural network via individual training data set are considered in the present research to carry out a series of experiments with different scenarios for the purpose of verifying the proposed techniques, obviously.