Conclusions & Further Researchs

Conclusions

First, wavelets are useful in image processing because they are able to represent functions that contain localized features. Other transform methods, like the Fourier transform, are not localized in space, therefore they cannot represent certain signal sets well. Second, using bitmap (bmp) file with black background color for signature images will give better performance in retrieving image. Third, compared to other wavelet, Biorthogonal provides better-embedded and retrieved images on data hiding.
At last, the highest Mean Square Error for variation in signature images is displayed by the black background color while for variation in wavelet the highest Mean Square Error is performed by Sym2 wavelet.

Further Researchs

Based on the above conclusion, it is important to investigate the wavelet-based data embedding including improvements of robustness for further studies.