2D Image Processing Layout
Filtering
We will use the same set of filters as for 1D signals, except that, on each image, two filters will be applied
- one is applied to all the rows in parallel
- the other one is applied to all the columns in parallel
Image structure
Since two filters are applied on each image, the image structure will have three fields
- the s field will hold the data
- the rd will be the delay cause by the filtering on the rows
- the cd will be the delay cause by the filtering on the columns
For instance, a grayscale image is initialized the following way:
grayImage = double(grayImage);
grayImage.s = grayImage;grayImage.rd=0;grayImage.cd=0;
Wavelet Transform structure
As in the 1D case, the wavelet transform WT will have two fields: WT.LoRes and WT.Details.
Similarly, WT.Details will be a column cell array
WT.Details = cell(scale,1)
The cell object WT.Details{i} will not be, however, an image object; it will be a structure with three fields. This is because three detail objects are produced in the processing:
- WT.Details{i}.r is the image object obtained by applying the band pass filter on the rows and the low pass on the columns
- WT.Details{i}.c is the image object obtained by applying the low pass filter on the rows and the band pass on the columns
- WT.Details{i}.d is the image object obtained by applying the band pass filter on the rows and on the columns