This paper shows how to detect the edge by a trous algorithm based on the quadric B-spline wavelet,divide the image into small blocks,choose appropriate edge detection threshold for each by human-robot interaction,and use edge tracking and compensating algorithm to improve the edge detection performance.

A trous algorithm solves the problem of the two sampling algorithm with Mallat and it has the feature that the translation is invariant and that each scale length of the data has the same as the original data s so as to make spectrum analysis on them.

The electrocardiogram was demarcated precisely by using the outstanding time-frequency focusing characteristics of wavelet transform,the time invariability of the à trous algorithm and the reference information from manual demarcation.

From the published papers,we find that most effort is paid to investigate the application of Mallat algorithm and à trous algorithm, some papers have related to registration.

Using Lacunaris algorithm the wavelet transform is programmed by programming language C, the correctness of the algorithm and program is verified by Microsoft Visual C++ 6.

Aiming at the boundary problem of wavelets transforms, an improved á trous algorithm was proposed.