提出了一种基于机器视觉技术的血袋编号字符识别方法, 建立了字符在线识别硬件与软件系统。运用经典标定算法确定摄像机内外参数。对扫描的原始图像进行了灰度化、增强、滤波、反相等预处理及字符的倾斜校正、图像行字切分、归一化操作。分析表明: 行字切分大于临界阈值T 时, 确定是文字像素一部分。预测结果与实测数据对比表明采用神经网络对转印字符进行识别可有效提高检测一次性血袋出厂编号的效率。相关技术及研制系统已应用于长春市某医疗制品企业, 取得较好识别效果及应用效果。

A method of blood bag numbering character recognition based on machine vision technology is proposed, and the hardware and software system of character on-line recognition are established. The internal and external parameters of the camera are determined by the classical calibration algorithm. The original images of the scanning are gray-scale, enhancement, filtering, anti-equal preprocessing, character tilt correction, image line segmentation and normalization operation.The analysis shows that when the line segmentation is greater than the critical threshold T, it is determined to be a part of the text pixel. The comparison between the predicted results and the measured data shows that the recognition of the transfer characters by neural network can effectively improve the efficiency of detecting the single blood bag factory number. The hardware and software technology and the developed system has been applied in Changchun some medical products enterprise, obtain good recognition effect and application effect.