具有领域独立性的通用模式识别方法的研究

Research on Domain-independent Universal Pattern Recognition

统计方法是当前广泛采用的模式识别方法,但是这种方法的专业性较强,缺乏通用性。针对这种现状文章提出了基于结构模式识别理论的一种通用模式识别方法,该方法使用一系列结构检测子来提取信号特征,具有领域独立性。设计了对比实验,通过同传统的统计模式识别方法的比较验证了通用模式识别方法的有效性,通过两个不同领域的模式识别问题验证了该方法的领域独立性和有效性。

Statistical method is widely used in pattern recognition. But it has strong specialty and is lacking for curreny. I propose the method of universal pattern recognition based on the structural pattern recognition theory. It employs a series of structure detectors to extract the signal features. It is independent in domain. The universal pattern recognition was compared with traditional statistical method by simulating experiment. It demonstrates this method is efficiency and verifies its domain-independent by processing the pattern recognition problem in two different fields.