摘要
研究了一种基于图像处理的档案盒标签检测技术,通过摄像头拍摄存放在档案柜窗口区的一排档案以采集图像,利用U-Net网络对档案盒标签图片进行分割,经仿射变换矫正图片裁剪得到单本档案盒标签图片,最后利用surf匹配算法对库中档案盒标签进行检测.实验结果表明,该方法测得的档案盒图像匹配正确率可达99.60%.
This article studies a detection technology of archive box labels based on image processing,which captures an image of labels of archive boxes in the window area of the cabinet by a camera in front of the desk of the cabinet,segments the image of labels by U-Net network,and then obtains a single archive box label picture corrected by affine transformation,and finally,uses the surf algorithm to match the picture with the database.The experimental results show that the matching accuracy rate of archive boxes measured by this method reaches up to 99.60%.
作者
李婧
苏叶
徐寅林
Li Jing;Su Ye;Xu Yinlin(School of Computer and Electronic Information,Nanjing Normal University,Nanjing 210023,China)
出处
《南京师范大学学报(工程技术版)》
CAS
2021年第2期60-64,共5页
Journal of Nanjing Normal University(Engineering and Technology Edition)
关键词
档案检测
图像分割
图像匹配
archive detection
image segmentation
image matching