We started a program of identification of faint structures in PNe on HST archive images, based on the “à trous” wavelet transform (Starck et al. 1998). Using this technique, an image is decomposed in a new set of N images, each with the same number of pixels as the original one, but containing only the structures of a characteristic scale size (going from one image to the next, the characteristic scale vary by a factor of two). The original image c0 can be expressed as the sum of all wavelet scales plus a residual smoothed array cN by C0(x, y) = CN(x, y) + Σj=1Nwj(x, y).