Pattern analysis on shift, rotation, and scaling

Numerous methods for recognising patterns invariant to their rotation, shift and scaling within an image have been proposed. These methods usually suffer from a lack of information as to the geometric feature estimation of tested patterns. In pattern analysis, this information is quite important. The article presents an improved algorithm applied to pattern classification and geometric feature estimation with good results.