The discrete wavelet transform can be used to process continuous-time signals. To use it, the initialization errors must be minimized. This is the aim of this paper. The results are justified and are presented in a unitary manner. We give a strategy to accomplish this minimization. Some examples are presented. A superior bound of these errors is also presented.
