摘要
针对数显仪表在复杂环境中的定位问题,对其颜色特征和边缘特征进行了研究。为了保证实时性和有效性,基于图像处理方式,提出了一种简单、有效的数显仪表定位方法。通过分析数码管颜色特征,得到数显式仪表的显著特征图;在特征图的基础上,使用分水岭算法和形态算子找到可能含有数码管的子区域。采用有效结合数码管颜色和边缘信息的算法,即大津阈值法和分水岭算法进行前景与背景的分离,在前景图像上作投影分割,以得到精确的数码管字符区域。采用局部二值模式特征和支持向量机分类的机器学习算法排除非字符区域,进一步提高定位准确率。试验结果表明,该方法可以在复杂环境下有效地定位数码管,能适应不同的光照和数码管大小、角度及颜色变化,且能够迁移到其他显示设备的定位,具有实际工程应用价值。
Aiming at the positioning issue of digital display instruments in complex environment,the color features and the edge features of digital instrument are studied. In order to ensure the real-time performance and effectiveness,based on image processing modes,a simple and effective positioning method for digital instruments is proposed. Through analyzing the color features of the digital tube,a prominent feature graphic of the digital display instrument is obtained; based on the feature graphic,the watershed algorithm and the shape operator are used to find the sub-regions that may contain the digital tube. By adopting the algorithm which effectively combines the color information and edge information of the digital tube,i. e.,the Otsu threshold algorithm and the watershed algorithm,the foreground and background are separated; then projection segmentation is performed on the foreground image to obtain an accurate character area of digital tube. The non-character area is excluded by local binary pattern feature extraction and support vector machine classification,thus the positioning accuracy can be enhanced. As experimental results show that the method proposed can locate the digital tubesin complex environment and adapt to different light,size,angle and color change,can be migrated to other display devices,with practical engineering application value.
作者
彭刚
杜兵
PENG Gang;DU Bing(School of Automation,Huazhong University of Science and Technology,Wuhan 430074,China;Key Laboratory of Image Processing and Intelligent Control of Education Ministry,Wuhan 430074,China)
出处
《自动化仪表》
CAS
2018年第7期56-60,共5页
Process Automation Instrumentation
基金
中国南方电网重点科技基金资助项目(GZKJQQ00000164)
关键词
数显仪表
分水岭算法
颜色特征
巡检系统
局部二值特征
投影分割
机器学习
Digital display instrument
Watershed algorithm
Color feature
Inspection system
Local binalT feature
Projection segmentation
Machine learning
