On the Theoretic Precision of Least Squares Matching Based on Features

This paper discusses a new approach of feature extraction used by SODAMS, the precision of least squares matching based on features, the signal-to-noise ratio of features and some thresholds used in feature detection. The analysis shows how precision depends on the gradient of the feature and on the direction of the edge line. The optimal precision of least squares matching based on features can be tess than 1μm.