摘要
借鉴统计模式识别方法,提出了基于动态边缘的OCR算法,以字符形态为分析样本,确定在动态视频图像中字符特征边缘区域的选择,然后找出每个字符的体征特性,通过大量试验得出每个字符的特征规律,最后通过相似性概率统计方法实现字符的识别。经在内蒙古超高压供电局航检视频处理中的实际应用,该算法准确率高达99.45%,为复杂背景下视频字符识别提供了新算法。
Put forward the OCR algorithm based on dynamic edge according to the statistical pattern recognition. Take the character shape as the sample of analysis, confirm the choice of the edge area for the character shape in the dynamic video. And then find out the signs characteristics of each character. Through a large amount of tests, gain the characteristic rules of each character. Finally through the method of probability statistics for similarity, realize the identification of the characters. After the practical application of aerial video processing in Inner Mongolia EHV Power Supply Bureau, the accuracy of the algorithm can reach 99.45%, which can provide a new algorithm for video character recognition under complex background.
作者
杨伟旗
关鑫
汪洋
皇甫云霞
YANG Weiqi;GUAN Xin;WANG Yang;HUANGFU Yunxia(Inner Mongolia EHV Power Supply Bureau, Hohhot 010080, China)
出处
《内蒙古电力技术》
2017年第6期16-20,共5页
Inner Mongolia Electric Power
基金
内蒙古电力(集团)有限责任公司2016年第一批科技计划项目(20161073)
关键词
OCR技术
航检视频
动态边缘
字符特征
统计模式识别
OCR technology
aeronautical detecting videos
dynamic edge
character feature
statistical pattern recognition
