衡器检测中数据自动识别采集方法研究

An automatic data identification and collection method for weighting instrument

为了减轻衡器检测过程中检测人员的工作量,提高检测质量,本文提出了一种基于液晶屏的数据识别采集方法,该方法首先读入衡器上含有液晶屏数字的图像,在手动标注出数字区域的基础上,通过中值滤波、二值化、形态学闭运算、数字分割、数字图像归一化等处理,得到待识别的单个数字图像,然后提取数字图像的水平方向和垂直方向的直方图作为特征,最后用K最近邻算法识别出各个数字。实验结果表明,该方法能够正确识别出衡器液晶屏中的数字。

In order to reduce the workload of weighting instrument inspectors and improve the detection quality, a digit recognition method for weighting instrument with liquid crystal screen is proposed. Firstly, image containing liquid crystal screen with numbers is loaded, and then, on the basis of manual labeling the digital region, the image is processed by a series of operation including median filtering, binarization, morphological closed operation, digit segmentation, digit image normalization and so on.Secondly, the histogram of the horizontal and vertical directions of the single digital image is extracted as feature, and the digits are recognized by k-Nearest Neighbor algorithm. The experimental results show that the proposed method can correctly recognize digits in the weighting instrument with liquid crystal screen.