摘要
针对实际环境下的数字仪表图像光照不均匀、产生畸变等问题,提出一种自适应的数字字符定位方法.首先从图像预处理入手,针对图像灰度不均匀,采用动态自适应阈值的S.D.Yanowitz二值化算法,能很好地分割目标和背景.然后再利用分级检测的Hough变换计算倾角进行图像矫正,最后利用投影算法进行逐步求精的字符精确定位.实验结果表明,该算法具有较好的鲁棒性,满足识别系统的要求.
A self-adaptive digital character location algorithm was proposed for the actual mechanic-digital meter image, which is not uniform in intensity and distorts sometime. First step was pre-processing： due to the nonuniform grey-scale, using S. D. Yanowitz dynamic self-adaptive threshold value binarization algorithm which can carve up object from the background. Then the inclined angle changed by grading test Hough transform algorithm was used to rectificate the image. Finally, a wise step of refinement strategy was adopted for digit character location by projection algorithm. Experiment results prove the algorithm has better robustness and meets the recognition system needs.
出处
《南京工业大学学报（自然科学版）》
CAS
2007年第4期93-97,共5页
Journal of Nanjing Tech University(Natural Science Edition)
关键词
二值化
图像矫正
字符定位
HOUGH变换
binarization
image rectification
character location
Hough transform
