基于倍频信号滤波的非规则图像动态特征提取  

Extraction of Non-regular Image Dynamic Characteristics Based on Frequency Doubling Signal Filtering

作　　者：张锦华[1] 孙挺[1] 

机构地区：[1]周口师范学院计算机科学与技术学院,河南周口466001

出　　处：《控制工程》2015年第2期301-305,共5页Control Engineering of China

基　　金：河南省科技厅软科学研究计划项目(132400410934);河南省科技厅科技发展计划科技攻关项目(122400450356)

摘　　要：传统方法采用小波动态特征提取方法,受到梯度向量运算阈值的干扰导致特征提取困难。提出一种基于倍频信号滤波的非规则图像动态特征提取方法。采用有限元方法进行图像湍流模型建模和流场分析,采用加速度传感器采集图像的冲击脉冲信号,采集的非规则图像动态特征信号输入后作乘法运算,输出结果为两者的乘积,再经低通滤波将倍频信号滤除,留下低频输出信号,设计低通倍频信号滤波器实现对噪声的滤波抑制处理,通过比较输出响应与期望响应产生估计误差,自动调整滤波器参数,最后进行动态特征提取算法改进。仿真表明,该算法能更加准确刻画非规则图像的动态特性,对图像运转的细节特征描述更加准确,特征提取稳健性较好,性能优越。Using wavelet dynamic feature extraction method, the traditional method is disturbed by gradient vector operation threshold, which leads to difficulty in characteristic extracting. This paper presents an approach to extract non-regular image dynamic characteristics based on frequency doubling signal filtering. By using the finite element method for analysis of image turbulence model and flow field, monitoring of image fault diagnosis, using the tooth surface of accelerometer shock for pulse signal acquisition, non-regular image dynamic characteristic signal is acquired as input. After multiplication, frequency signal in output is filtered through the low-pass filter, staying low frequency output signal. The low-pass frequency signal processing filter is designed for noise filtering and suppressing. By comparing the output response and the desired response, estimation errors are generated for the automatic adjusting of the filter parameters, and finally the dynamic characteristic algorithm is improved. Simulation results show that the algorithm has excellent performance and can describe the dynamic characteristics and detailed features of non-regular image more accurately and extract features more stably.

关 键 词：图像 状态监测 故障诊断 信号滤波 

分 类 号：V240.2[航空宇航科学与技术—飞行器设计]