雷达态势图像表格检测与识别

Table Detection and Recognition of Radar Situation Image

随着军贸出口的指挥控制类武器装备种类增多,设计兼容不同国家、不同体制的雷达接口非常关键。而如何高效获取接口不开放的雷达目标信息是核心问题之一。因此,设计了一种基于数学形态学和深度学习的图像识别方法,用于获取雷达图像的情报信息,实现雷达图像信息的数字化转换,并将数据上传至指挥中心。通过网络摄像头实时采集雷达态势图像信息,采用数学形态学方法实现雷达图像文本定位与检测,并采用CRNN深度学习模型,完成雷达目标情报的识别与提取。结果表明,平均每帧雷达图像的识别时间在500 ms以内,准确率可以达到95%以上,满足实时性和准确性的要求。

With the increase of the types of command and control weapons and equipment exported in military trade,the design of radar interfaces compatible with different countries and systems has become a key technology.How to efficiently obtain radar target information with closed interface is one of the core problems it faces.In this paper,an image recognition method based on mathematical morphology and deep learning is designed to obtain the intelligence information of radar image,realize the digital conversion of radar image information,and upload the data to the command center.The radar situation image information is collected by webcam in real time,while the text positioning and detection of radar image is realized by mathematical morphology method.Finally,the recognition and extraction of radar target information is completed by CRNN deep learning model.The results show that the average recognition time of each frame of radar image is less than 500 ms,and that the accuracy can reach more than 95%,which meets the requirements of real-time and accuracy.