【摘要】 在研制万能工具显微镜(UTM)自动读数系统的过程中,针对UTM目镜视场数字字符的图像特点,提出了一种数字字符识别预处理算法。算法利用Roberts边缘检测算子对图像进行二值化处理,设计了数学形态学算子进行除噪运算,使用了一种改进的投影算法进行字符分割。结合模板匹配算法的识别实验得到的识别率为98.9%。更多还原

【Abstract】 A pretreatment algorithm of numeral character recognition was presented for self-reading system of Universal Tools Microscope(UTM),which was based on the characteristics of numeral characters in the view field of eye lens on UTM.The character area was segmented by locating millimeter line in the algorithm.After analysis on several experimental results,the Roberts operator wasemployed for edge detection;the image was binarizated according to the threshold of gradient,and the noise of Spiral line in the background of the image was eliminated effectively.Furthermore,two mathematical morphology operators were constructed to filter the noise in single-step processing.In the character segmentation,a projective algorithm based on characters′features was adopted.The experiments with template matching algorithm were performed,and the results show that the recognition rate is 98.9%.更多还原