摘要
由于设备异常检测过程的复杂性与技术局限性,传统人工方法进行设备异常检测时,由于没有故障样本,通常需要耗费大量人力成本和财力成本,不能解决设备本地资源的有限性问题,并且采集到的设备异常部位图像可辨识程度不高,不能对检测数据进行可视化处理,不能保证异常检测结果的可信度,进而无法上传到设备检测中心以图像的方式留存系统信息,因此无法获得满意的检测效果。为此,提出基于智能化图像处理的设备异常检测方法。针对采集的原始设备故障识别度不高的缺陷,通过图像灰度化处理、灰度拉伸和中值滤波等方法对图像进行预处理,提高设备异常检测的图像质量,为图像检索和提高设备异常检测效率提供技术支持,依据相关原理进行图像特征分解提取,并结合一定的匹配机制对图像进行相似度匹配,根据实际检测需求进行特征权重调整,获得较为准确的检测结果,以实现设备的有效异常检测。实验结果表明,采用改进算法进行设备的异常检测,能够提高设备检测系统运行的效率保证检测的准确性,满足了设备检测自动化的实际需求。
Because of the complexity of the anomaly detection process equipment and technical limitations, the traditional manual method of anomaly detection equipment, because of the lack of fault samples usually requires a lot of human and financial costs, can not solve the problem of equipment limited local resources, and collected equip- ment abnormal spot image recognition degree is not high, not to treat data visualization, cannot ensure the credibility of the anomaly detection results, it can no longer be uploaded to the equipment testing center system information is re- tained in the form of image, so we can not get satisfactory result of testing. For this, put forward equipment anomaly detection method based on intelligent image processing. For acquisition of the original defects of equipment fault rec- ognition degree is not high, through the image gray processing, gray stretch and median filtering method for image preprocessing, image quality, to improve equipment anomaly detection for image retrieval, and improve the efficiency of anomaly detection equipment to provide technical support, extract image feature decomposition according to the rel- evant principles, and combined with certain matching mechanism for image similarity matching, according to the ac- tual testing requirements in terms of feature weight adjustment, to obtain more accurate test results, in order to realize effective anomaly detection equipment. The experimental results show that the improved algorithm of anomaly detec- tion equipment, to improve the efficiency of the detection system operation, ensure the accuracy of the detection to meet the needs of the test automation equipment.
出处
《计算机仿真》
CSCD
北大核心
2016年第1期425-429,共5页
Computer Simulation
基金
吉林省科技厅项目(20130102030JC)
关键词
图像处理
设备检测
特征提取
相似度匹配
Image processing
Equipment inspection
Feature extraction
Similarity matching
