基于传感器阵列的SF_(6)仪表图像识别方法  

The SF_(6)Instrument Image Recognition Method Based on the Sensor Array

作　　者：赵炜 饶雪梅 谢榕 李杰 刘鑫 李汉豪 ZHAO Wei;RAO Xuemei;XIE Rong;LI Jie;LIU Xin;LI Hanhao(Bijie Power Supply Bureau of Guizhou Power Grid Co.,Ltd.,Bijie 563200,China;Guangzhou Andian Measurement and Control Technology Co.,Ltd.,Guangzhou 510000,China)

机构地区：[1]贵州电网有限责任公司毕节供电局,贵州毕节563200 [2]广州安电测控技术有限公司,广东广州510000

出　　处：《传感器世界》2024年第8期35-40,共6页Sensor World

基　　金：2023年指针式(数字式)表计图像数据智能识别研究项目(No.GZDW-FW-0490-22-01)。

摘　　要：SF_(6)仪表图像中存在多个亮暗区域、反光和阴影等问题,以单一传感器为基础的识别方法缺陷明显。为此,引入传感器阵列,提出SF_(6)仪表图像识别方法。将索尼的IMX219作为核心图像传感器,设计传感器阵列,全方位捕捉SF_(6)仪表图像信息,在TCP/IP协议下实时处理和传输传感器阵列采集的数据。通过特征向量和相似度参数对图像去噪与多尺度小波变换,得到不同尺度下的纹理特征。利用支持向量机(Support Vector Machine,SVM)实现对SF_(6)仪表图像信息的识别。测试结果表明,该设计图像识别方法不仅具有优越的去噪效果,对于具体示数的识别结果也有更低的误差。SF_(6)instrument image light dark area,there are multiple reflections and shadows,defect recognition method based on single sensor is obvious.Therefore,the sensor array is introduced and the image recognition method of SF_(6)instrument is proposed.Using SONY's IMX219 as the core image sensor,the sensor array is designed to capture the SF_(6)instrument image information in all aspects,and the data collected by the sensor array is processed and transmitted in real time under TCP/IP protocol.Image denoising and multi-scale wavelet transform are used to obtain the texture features at different scales through feature vectors and similarity parameters.Support vector machine(SVM)recognition was used to recognize SF_(6)instrument image information.In the test results,the image recognition method designed in this paper not only has superior denoising effect,but also has lower error for the recognition results of specific symbols.

关 键 词：传感器阵列 SF_(6)仪表 信息识别 索尼IMX219 图像传感器 图像去噪 

分 类 号：TH83[机械工程—仪器科学与技术]