I know the three Haar basis functions
//mother wavelet:
[1 1 -1 -1]
//dilation:
[1 -1 0 0]
//translation:
[0 0 1 -1]
However, the Haar basis consists of four basis elements: The mother wavelet, the translation, the dilation and another, fourth element, which I don't understand. According to this, this fourth element is
[1 1 1 1]
But how would this one look like, and how is it obtained (dilation/translation, whatever combination)? This confuses me.
Btw: is the Haar basis defined as
[ 1 1 1 1
1 1 -1 -1
1 -1 0 0
0 0 1 -1]
or as the transpose of this?
If I understand the notation right, it looks like the [1 1 1 1] element is just the DC coefficient, i.e. the mean of the input signal.
By symmetry, the Haar wavelet integrates to 0 over its domain. Thus, if you didn't include the DC coefficient, adding a constant bias to the input signal would not change its Haar transform at all.