一种字符编码标志识别定位方法

Recognition and location method for the character coded marks

针对在大尺寸结构动态摄影高精度、自动化测量中如何有效提高人工标志的识别效率和准确率等问题,该文提出一种新的字符编码标志设计及其识别定位方法。该标志由两部分组成,实心圆部分取背景色,字符部分取前景色且两者的灰度具有显著差异,是一种由字符确定码值的圆形编码标志。采用多级阈值进行特征分割,分别得到字符区域和圆形区域;生成编码字符样本,并采用BP神经网络训练,进而实现精确的分类识别解码;对背景圆区域内部进行填充修复和灰度平滑,采用灰度质心法可实现亚像素定位。该类型编码标志设计简单且提出的字符识别方法解码快速准确。

Aiming at the problem of how to improve the recognition efficiency and accuracy of artificial marks was effectively in the high-precision and automatic measurement of large-size structure dynamic photography, a novel character-coded mark design is proposed. Then, the recognition and location methods are researched. This kind of mark is composed of two parts. The solid circle part has the background color and the character part has the foreground color. The gray level of the two parts is significantly different. It is a circular coded mark and the code value is determined by the character. Multi-level thresholds are used to get character region and circular region respectively. Character samples are generated and trained by back propagation neural network to achieve accurate classification, recognition and decoding. The inner of the background circle region is filled and repaired by gray level smoothing. Sub-pixel location can be achieved by gray centroid method. The design of the coded mark is simple. The accuracy of character recognition is reliable.