基于CTPN与Tesseract的机载雷达视频字符识别  

Character Recognition for Airborne Radar Video Based on CTPN and Tesseract

作　　者：邢宝峻 彭晓明[1] 王卫星 XING Bao-jun;PENG Xiao-ming;WANG Wei-xing(Air Force Early Warning Academy,Wuhan 430019,China)

机构地区：[1]空军预警学院,湖北武汉430019

出　　处：《舰船电子对抗》2022年第1期81-84,共4页Shipboard Electronic Countermeasure

摘　　要：为实现对机载雷达视频的识别,正确评估飞行员的机载雷达操纵水平,提出了一种基于连接文本提议网络(CTPN)与Tesseract相结合的对机载雷达视频中关键字符进行识别的方法。通过视频压缩、图像截取、图像拼接、腐蚀与膨胀等操作,对机载雷达视频进行预处理,接着将视频转化为图像。通过CTPN对图像进行文本检测,最后调用Tesseract对检测出的字符进行识别,形成文本输出,实现对机载雷达视频的自动化识别,从而替代现有的人工判读方式,达到提升机载雷达视频分析效率的目的,为借助视频分析快速准确评估飞行员的雷达操纵水平提供依据。In order to realize the recognition of airborne radar video and correctly evaluate the pilot’s control level of airborne radar,this paper proposes a method based on the combination of connec-tionist text proposal network(CTPN)and Tesseract to recognize the key characters in airborne ra-dar video.This paper preprocesses the airborne radar video through operations such as video com-pression,image interception,image splicing,corrosion and expansion,etc.,and then converts the video signals into image signals,performs text detection to the image through CTPN,finally calls Tesseract to recognize the detected characters and form text output,realizes automatic recognition of airborne radar video,thereby replaces the existing manual interpretation mode,achieves the pur-pose of improving the efficiency of airborne radar video analysis,provides a basis for quickly and accurately assessing the pilot’s control level of airborne radar with the help of video analysis.

关 键 词：机载雷达视频 文本检测 字符识别 

分 类 号：TN957.52[电子电信—信号与信息处理]