There are many literatures on pattern recognition by the use of the method for the cross correlation when an unknown pattern is classified in the pattern recognition. But, this method for the pattern recognition possesses the next two weak points. (1) The typical pattern and the unknown pattern must be fitted at the center of the gravity with each pattern and then the cross correlation of the typical pattern and the unknown pattern must be taken. (2) The size of the unknown pattern must be standardized. As the result of these weak points, the unknown pattern is restricted in the position and the size. Then this paper shows the improved pattern recognition by the use of the method for the cross correlation, where the given pattern is lowered in dimension by a pre-procedure and reformed unknown pattern newly to eliminate the preceding weak points.