隐马尔柯夫模型的一种面向对象的程序实现框架

在以Ｃ＋ ＋ 语言作为程序设计语言的基础上，提出了一种用面向对象的程序设计方法来实现隐马尔柯夫模型的基本框架。按照这一框架实现的隐马尔柯夫模型能够很好地保护它自己不受未知的外部事件的影响，从而使它的数据和功能免遭破坏；而且对模型的改进以及程序实现上的变化也不会影响使用它的外部应用程序。此外，用这一方法编制的隐马尔柯夫模型程序具有很好的重用性和继承性。

An Object-Oriented Programming Approach to Implement Hidden Markov Model

Hidden Markov model (HMM) is an important stochastic model of signal and has a wide range of applications. But HMM is too complicated to be implemented efficiently by conventional programming methods. Here we present an object oriented programming (OOP) approach to implement HMM by encapsulating it in a class using C as the programming language. Such an approach, to our best knowledge, has not been reported in the open literatures. In Section 1, we define five data types in the class representing the five elements of HMM. In Section 2, we first employ forward backward algorithm (eqs.1 through 5), Viterbi algorithm (eqs.6 through 12), and Baum Welch algorithm (eqs.13 through 17) to solve three basic problems of HMM. We then define five functions in the class which compute the variables used in the above mentioned three algorithms. Based on the five functions, we construct three functions to realize three algorithms. This approach has already been used in the development of a speaker dependent isolated word recognition system, and has met with preliminary success.