基于Relief算法的激光干涉仪故障图像自动识别方法研究  

Research on automatic recognition method of laser interferometer fault image based on relief algorithm

作　　者：李嘉恩 LI Jiaen(GuangDong NanFang Institute of Technology,Jiangmen Guangdong 529040,China)

机构地区：[1]广东南方职业学院,广东江门529040

出　　处：《自动化与仪器仪表》2020年第12期29-32,共4页Automation & Instrumentation

基　　金：广东省自然科学基金项目(No.04011761)。

摘　　要：传统激光干涉仪故障图像识别方法将提取的所有故障图像特征作为输入进行故障图像识别,计算量大,且影响识别结果的准确性。为此,提出一种基于Relief算法的激光干涉仪故障图像自动识别方法。计算灰度共生矩阵,依据灰度共生矩阵求解获取纹理特征参数,构成特征子集对图像纹理特征进行描述。利用Relief算法对实例集合任意采样,求出各属性的权重对特征进行选择,从而实现对特征的排序,得到对识别作用最大的前几个特征,把特征权向量和原始样本数据共同输入相关向量机中进行训练,建立故障图像识别分类器,自动实现激光干涉仪故障图像的自动识别。经验证,Relief算法选择的特征可有效区分故障样本与非故障样本,所提方法对故障图像的识别准确性高。the traditional laser interferometer fault image recognition method uses all the fault image features extracted as input for fault image recognition,which requires a lot of calculation and affects the accuracy of the recognition results.Therefore,an automatic recognition method of laser interferometer fault image based on relief algorithm is proposed.The gray level co-occurrence matrix is calculated,and the texture feature parameters are obtained according to the gray level co-occurrence matrix,and the feature subsets are formed to describe the image texture features.By using relief algorithm to sample the sample set arbitrarily,the weight of each attribute is calculated to select the feature,so as to realize the sorting of the feature,and the first features that have the greatest effect on the recognition are obtained.The feature weight vector and the original sample data are input into the correlation vector machine for training,and the fault image recognition classifier is established to realize the automatic recognition of the fault image of the laser interferometer No.It is verified that the features selected by relief algorithm can effectively distinguish fault samples from non fault samples,and the proposed method has high accuracy in fault image recognition.

关 键 词：RELIEF算法 激光干涉仪 故障图像 识别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]