Mahalanobis Distance itself is just a mathematical calculation called generalized variance. The essential problem is a set of original data to construct Mahalanobis Space.
A new method for pattern recognition for hand writing letter is proposed in this article. The method is to construct Mahalanobis Space by using number of lines and the range of lines in each row and column. In the case of 32×32 cells in a square, we have 1024 digital response numbers. So, it is almost impossible to construct Mahalanobis space as we have to deal with 1024×1024 matrics. The method here needs to treat only smaller than 124×124 matrics. The method stated in this article may be also applied for many other pattern recognition as seen in the next article.