【作者】 郝如江； 冯志鹏； 褚福磊；

【Author】 Hao Rujiang~(1,2) Feng Zhipeng~3 Chu Fulei~1 (1 Department of Precision Instruments and Mechanology, Tsinghua University, Beijing, 100084, China; 2 Department of Computer and Information Engineering, Shijiazhuang Railway Institute, 050043, China; 3 Research Institute of Vehicle Engineering, University of Science and Technology Beijing, Beijing, 100083, China)

【机构】 清华大学精密仪器与机械学系； 石家庄铁道学院计算机与信息工程分院； 北京科技大学车辆工程研究所；

【摘要】 分析了形态滤波算子的构造及结构元素的选取原则,设计了多尺度混合形态滤波器用于滤除噪声及与分析信号无关的谐波成分,然后再进行多尺度形态闭开的差值滤波处理,得到了明显的特征信息。通过对仿真数据和实际信号的分析表明,设计合适的形态滤波器既可以抑制噪声又能够提取特征信息,形态滤波算法灵活简单,适于实时的信号处理和特征分析。更多还原

【Abstract】 Morphological filter is a no-linear filtering method based on mathematical morphology.In this paper, how to design the morphological operators and select the structuring elements are analyzed.The multi-scale morphological filter with mixed operators is devised to remove the noise and other harmonic components.Then, the morphological closing-opening difference filter is used to extract the feature components.The simulated data and the experimental signals are processed by the morphological filters,the analysis results indicate that the morphological filters with proper operators and structuring elements can not only remove noises but also extract the useful characteristic information in the signals.The morphological filters are suitable for signal analysis on-line considering the simple and flexible arithmetic.更多还原