摘要
针对指针式仪表的判读和计量校准,本文提出了一种新颖实用的基于数字图像处理技术的读数识别方法。首先对图像进行预处理,包括图像分割、图像滤波及膨胀细化等步骤,其中采用了一种改进的自适应中值滤波方法来改善图像效果;然后用改进的、双阈值的霍夫变换提取出图像特征信息,创新性地利用仪表表盘刻度与角度的线性关系得到指针读数,并且对于指针与整刻度线重合的情况进行了后期校正。实验结果表明,该方法对于指针角度识别十分有效,且具有较高的精度和实时性,有很强的实际应用价值。
As to the identification and calibration of index-instrument,a novel method for reading recognition based on digital image processing is proposed.Firstly,preprocessing operation on images is done,including image segmentation,image filtering,image dilation and thinning etc,and a method of improved adaptive median filter is adopted to improve the image.Then,characteristic information of image is extracted with the improved dual thresholds Hough transformation,and the pointer reading is got using the linear relationship between scale and angle of instrument dial revolutionary.And as to the situation that pointer coincides with the whole tick,late correction has been done.The experiment results show that the algorithm is very valid for the pointer angle recognition,and has higher precision and real time.It has strong practical application value.
出处
《光电工程》
CAS
CSCD
北大核心
2011年第4期145-150,共6页
Opto-Electronic Engineering
基金
国家自然科学基金(60904004)
成都市科技计划(10GGYB527GX-023)
关键词
指针式仪表
读数识别
霍夫变换
index-instrument
reading recognition
Hough transformation
