FA中字符分割提取方法的研究

Studys on Character's Segmentation and Extract in FA

在ＦＡ中，由于环境光照的不均匀、现场的复杂性及产品位置、方向的变化性，使得提取产品表面的字符变得非常困难。文章在综合考虑生产实时过程的前提条件下，提出了一种将简单统计法与Ｒｏｂｅｒｔｓ边缘检测相结合的图象二值化方法，在此基础上，运用Ｈ－Ｓ连通分析方法，将非字符的背景去除，从而保留下待识别的字符串。为了使识别与字符的位置和旋转方向无关，利用多字符目标形心不变原理和Ｋ－Ｌ变换，对字符串的位置和方向进行了校正。最后利用投影值将单个字符分隔出来，为进一步的识别奠定了基础。

In the Factory Automation,it becomes very difficult to extract the characters from the surface of the products because of the non-uniformity of the environment light,the complexity of the spot,the location of the products and the variety of the direction. Based on the comprehensive consideration of the precondition of the real-time course for production,this paper raises the method for image's binary which combine the simple statistic method and Roberts edge detection method,and removes the non-character background as to maintain the string to be identified by the way of H-S connectivity analysis way. In order to make it no-relate with the location of the string and the rotation direction, the K-L transformation and the theory of invariable center of the shape for multi-character object are used to adjust the string location and direction. Then separate the character from another by the value of projection to make the foun dation of further identification.