This paper deals with the noise reduction of discrete AFM surface images using orthogonal wavelets. More in detail, it compares the usefulness of the Daubechies wavelets with different vanishing moments for noise reduction. The work is based upon the discrete wavelet transform (DWT) version of wavelet package transform (WPT). With the help of a seminorm the measurement of noise of a sequence is defined. An algorithm for noise reduction is proposed to detect unavoidable measured noise in topographic surface scans. The denoising wavelet algorithm is used to improve the quality of the scanned images. By taking real measurements in which the measurement and system noise is included, the effectiveness of the proposed denoising algorithm is validated.