摘要
针对带表格的标签图像中,表格与字符产生交叠的问题,本文设计了一种带表格的标签字符识别的预处理算法。采用Hough变换的方法对图像进行倾斜矫正;粗定位与精确定位相结合的两步定位方法对带表格的字符区域进行准确定位,在去除表格框线后恢复出框线和笔画的原始交叠区域,定位到带遮挡的字符区域;用垂直投影的方法对字符进行分割。实验结果证明,该算法能够应用于标签字符识别系统中,可很好的满足系统对实时性和稳定性的要求。
Form tags associated with the image,table and character overlapping problems, in this paper,we design a label character recognition preprocessing algorithm with form. Adopt the method of Hough transform for tilt correction;Combining coarse location and accurate positioning method of the positioning of the two steps with form accurate positioning, character areas in which the frame line removal form after recovering the original overlapping area of the frame lines and strokes,and character positioning to take shade,By the method of vertical projection character segmentation.The experimental results show that the algorithm can be applied to the TAB character recognition system,it can well meet the system requirements for real-time and stability.
关键词
表格字符
预处理算法
倾斜矫正
两步定位
垂直投影
Form character
The pretreatment algorithm
Tilt correction
Two step positioning
Vertical projection
