一种鲁棒的数显式仪表读数检测识别方法
|
摘要
本文提出了一种实用于巡检现场的数显式仪表读数检测识别方法。为了适应不同样式数显式仪表、不同光照情况下、不同拍摄角度获得的图片,先设计基于区域连通特性的数字区域粗定位算法,从复杂的背景中检测出候选数字区域。然后使用基于骨架特征的笔画宽度信息统计方法滤除大部分不符合笔画特征的候选区域,最后设计一系列基于形状特征、颜色一致性特征、排列特征的过滤器,进一步滤除所有非数字区域,实现复杂仪表盘上数字区域的准确定位。对获得的数字区域进行倾斜校正后送入Tesseract OCR引擎即可识别得到正确读数。实验结果表明该方法具有较高的准确度及较强的鲁棒性,处理速度快,可应用于复杂仪表柜的巡检。
An automatic detection and identification method for digital meters is proposed. Considering images taken from different styles of meters in different illumination conditions and varied camera angle, a coarse location method of digit regions based on region connectivity is first designed to find candidate digit regions from complicated backgrounds. Then a statistical method of stroke width information based on skeleton feature is designed to remove lots of the candidate regions which do not conform to stroke features. Finally, some filters based on digit shape feature, color coherence feature and permutation feature are designed to remove all other wrong candidate regions. After tilt correction, the detected regions are send to the Tesseract OCR engine to recognize the readings. The experimental results show that the method has high accuracy and strong robustness with high process speed. It can be used for inspection of complicated meter cabinets.
An automatic detection and identification method for digital meters is proposed. Considering images taken from different styles of meters in different illumination conditions and varied camera angle, a coarse location method of digit regions based on region connectivity is first designed to find candidate digit regions from complicated backgrounds. Then a statistical method of stroke width information based on skeleton feature is designed to remove lots of the candidate regions which do not conform to stroke features. Finally, some filters based on digit shape feature, color coherence feature and permutation feature are designed to remove all other wrong candidate regions. After tilt correction, the detected regions are send to the Tesseract OCR engine to recognize the readings. The experimental results show that the method has high accuracy and strong robustness with high process speed. It can be used for inspection of complicated meter cabinets.
引文
[1]BAI Q,ZHAO L,ZHANG Y,et al.Research of automatic recognition of digital meter reading based on intelligent image processing[C].2010 2nd International Conference on Computer Engineering and Technology(ICCET),IEEE,2010,5.
[2]张黎,刘争鸣,唐军.基于BP神经网络的手写数字识别方法的实现[J].自动化与仪器仪表,2015,(6):169-170.
[3]黄青丹,张亚茹,裴利强等.基于图像识别的仪表自动检定系统[J].测控技术,2016,35(10):153-156.
[4]王舒憬,杜皎洁.基于OPENCV的数字万用表数字识别方法[J].自动化与仪器仪表,2014,(1):62-63.
[5]支馨悦.基于安卓手机的液晶屏数字识别系统的设计与实现[J].自动化与仪器仪表,2016,(3):13-14.
[6]SMITH R.An overview of the Tesseract OCRengine[C].2007 Ninth International Conference on Document Analysis and Recognition(ICDAR).IEEE,2007,2.
[7]ZHANG H G,ZHAO K,et al.Text extraction from natural scene image:A survey[J].Neurocomputing,2013,122(12):310-323.
[8]CHEN X AND YUILLE A L.Detecting and reading text in natural scenes[C].IEEE Computer Society Conference on Computer Vision and Pattern Recognition(CVPR),2004.
[9]MATAS J,CHUM O,URBAN M,et al.Robust widebaseline stereo from maximally stable extremal regions[J].Image and vision computing,2004,22(10):761-767.
[10]EPSHTEIN B,OFEK E,WEXLER Y.Detecting text in natural scenes with stroke width transform[C].Proceedings of 2010 IEEE Conference on the Computer Vision and Pattern recognition(CVPR),2010.
[11]苑玮琦,江海燕.基于ACSS角点定位的液晶显示仪表示值识别算法[J].计算机应用系统,2015,24(7):111-116.
[12]袁俊淼.基于几何约束的笔划宽度变换(SWT)算法及其字幕文本定位应用[D].成都:电子科技大学,2015.
[13]刁智华,吴贝贝,毋媛媛等.基于图像处理的骨架提取算法的应用研究[J].计算机科学,2016,43(6A):232-235.
[14]王希晨.基于Tesseract的电子票据云平台的设计与实现[D].南京:南京大学,2016.
[2]张黎,刘争鸣,唐军.基于BP神经网络的手写数字识别方法的实现[J].自动化与仪器仪表,2015,(6):169-170.
[3]黄青丹,张亚茹,裴利强等.基于图像识别的仪表自动检定系统[J].测控技术,2016,35(10):153-156.
[4]王舒憬,杜皎洁.基于OPENCV的数字万用表数字识别方法[J].自动化与仪器仪表,2014,(1):62-63.
[5]支馨悦.基于安卓手机的液晶屏数字识别系统的设计与实现[J].自动化与仪器仪表,2016,(3):13-14.
[6]SMITH R.An overview of the Tesseract OCRengine[C].2007 Ninth International Conference on Document Analysis and Recognition(ICDAR).IEEE,2007,2.
[7]ZHANG H G,ZHAO K,et al.Text extraction from natural scene image:A survey[J].Neurocomputing,2013,122(12):310-323.
[8]CHEN X AND YUILLE A L.Detecting and reading text in natural scenes[C].IEEE Computer Society Conference on Computer Vision and Pattern Recognition(CVPR),2004.
[9]MATAS J,CHUM O,URBAN M,et al.Robust widebaseline stereo from maximally stable extremal regions[J].Image and vision computing,2004,22(10):761-767.
[10]EPSHTEIN B,OFEK E,WEXLER Y.Detecting text in natural scenes with stroke width transform[C].Proceedings of 2010 IEEE Conference on the Computer Vision and Pattern recognition(CVPR),2010.
[11]苑玮琦,江海燕.基于ACSS角点定位的液晶显示仪表示值识别算法[J].计算机应用系统,2015,24(7):111-116.
[12]袁俊淼.基于几何约束的笔划宽度变换(SWT)算法及其字幕文本定位应用[D].成都:电子科技大学,2015.
[13]刁智华,吴贝贝,毋媛媛等.基于图像处理的骨架提取算法的应用研究[J].计算机科学,2016,43(6A):232-235.
[14]王希晨.基于Tesseract的电子票据云平台的设计与实现[D].南京:南京大学,2016.