ABSTRACT

The E-Sarn Dharma Characters are not the standards printing of computer.  So, it is very difficult to build up an acceptable algorithm for recognizing them. Because of this, it is of several formalities and flexibilities for writing in other regions. And in the region in which the writer is living. So, in recognition and rememorize it must depend on the prevention f special characteristic from the distinguished characteristics. Thus, in order to be able to explain the importantly special characteristics of each alphabet mostly completely. In this research, we are specific to study and remember the 91 Dhamma Alphabets in each of which is of six forms E-Sarn (Northeastern Provinces).

 This study is of the usage of the method to adjust the size of the pixel of the alphabets in the form of non-linear in corporation with a number of the distinctive pattern of E-Sarn Damma Characters to determine the features for the recognition of algorism which utilizes the Hidden Markove Models (HMM), From the comparative study of the effectiveness of the E-Sarn Dhamma Alphabets with the numbers of the statistics of the patterns of the E-Sarn Dhamma Alphabets in the a definite and non-definite characteristics, it is found that the statistical numbers of E-Sarn Dhamma Alphabets are divided into two parts; the proper part and the least one which give the result at the rate of knowing and recognizing at the overate point of 98.9 percent which can be divided at the highest level for the patterns of E-Sarn Dhamma Alphabets of the left-right types, and of the numbers of the E-Sarn Dhamma Alphabets of between 20-40.