Investigating Digital Watermark Dynamics on Carrier File by Feed-Forward Neural Network

Carrier files are commonly described as host files in digital watermarking in which hidden files are embedded on it. As a result, new files are formed which contain the hidden files or messages. This paper aim at resolving the problem of capacity in Image watermarking and utilizes the bits ratios of the watermark and carrier file as the raw data for analysis. The data are obtained from the result of the first project undertaken to determine the implementation of different applications available in the public domain for embedding a watermark. Feed-forward neural network (FFNN) is used for analysis because is applicable to a wide range of forecasting problems and yields a high degree of accuracy for the bits ratios of watermark and host. The result indicates the relationship between the carrier file and the hidden file, which establishes a pattern where the larger the bits of the carrier file, the larger the watermark bits and vice versa. Although this is only in terms of Image watermarking. Further studies should apply the same technique on video and audio watermarking.