表示是将客观数据或现象利用数值或者编码方式进行描述,从而使其在一定数学含义框架下彼此相关的过程。在模式识别领域中,表示问题至关重要,其效果将直接影响后续步骤的复杂程度与分类性能。本文主要对表示在模式识别中的作用、表示问题的理解、表示的原则进行分析并对表示的方法进行归类描述,其目的是吸引更多的同行开展深入的表示问题研究,促进模式识别理论和方法的进步。
The representation is an important issue in pattern recognition, which describes a real world observation as numerical value or encodes and makes the observations relative in mathematical framework. The effect of representation influences the complexity and performance of post-processing directly.Representations' situation in pattern recognition, the comprehension and formula of the representation and method for typical representations are disscussed in this paper. It is concluded that although there are various method for representation, it is still an issue should be improved and studied.