Abstract: The main difficulty of recognizing meter parameters symbols is about locating and segmenting them from all the meter parameters.This paper uses image processing and recognition method and presents a new algorithm based on the symbols combined features of pixel density，characters center distance，connectivity and symmetry to segment main meter parameters symbols.The methods of unchanged moment feature and support vector machine are used to recognize parameters symbols.Experimental results show the combined features method works very well.

摘要： 仪表表盘参数符号识别的主要难点在于从仪表表盘中定位并切分出所需要的参数符号。采用图像处理和识别的方法，提出了基于字符的像素密度特征、中心点间距、连通域特征、对称性特征等综合特征，从拍摄的指针式仪表表盘图片中提取并分割出仪表盘的主要参数符号；再利用不变矩特征和支持向量机方法进行仪表表盘参数的识别。测试结果表明，这种综合特征方法实用有效。