Folks,
I have read a number of articles on Discrete Wavelet Transform (DWT) and looked at some sample code as well. However, I am not clear on what exactly does DWT achieve.
Here is what I understand. For a two dimensional image in YUV format, I can pass in the Y plane (brightness) to DWT function as a parameter. The function returns me a matrix of the original width and height containing coefficient values.
What are these coefficient values telling me? Is it how fast or slow the brightness of a pixel has changed compared to its neighbors?
Further, the returned matrix is rearranged in four quarters. As the coefficients have been rearranged, I no longer know which coefficient belongs to which pixel. This is confusing. If I cannot associate the coefficient to its corresponding pixel location, how can I really use the coefficients?
A little bit of background. I am looking at hiding some information in an image as an invisible watermark. From what I understand, DWT can help me identify the best region to hide the information. However, I have not been able to put the whole picture together.
Ok. I figured out how DWT works. I was under the assumption that the coefficients generated have a relationship with the original image. However, the transform converts the input luma into a completely different set. It is possible to run the reverse transform on the new values to once again obtain the original values.
Regards,
Peter