Good edge extraction from temporal series of solar images is fundamental in the solar astrolabe experiment. Noise and spots in images cause however, difficulties to extract an accurate solar edge. We present in this paper a new image-processing method which solves this problem and thus improves the characteristics of the solar astrolabe experiment. The method is based upon the use of wavelet transform in the solar image analysis. It is developed to remove image defects (parasite spots) and noise without reducing image resolution. Solar images obtained at Calern Observatory astrolabe (France) are then processed using this method. Solar edges of these images are extracted and trajectories reconstructed.