摘要
针对集装箱编号图像存在光照不均、背景复杂、模糊、污损、断裂等问题,本文提出,先采用基于深度学习的目标检测算法(YOLOv4)来实现集装箱编号区域定位,接着对定位后的集装箱编号图像进行预处理,采用连通域分割法分割字符,然后把一个个的字符送入模板匹配算法中进行字符识别。通过理论分析以及实验证明了本文方法的有效性,识别准确率相比其他方法明显提高。
In order to solve the problems of uneven illumination,complex background,blur,stain,fracture and so on in the container number images,this paper firstly uses the object detection algorithm based on deep learning(YOLOV4)to locate the container number area,preprocesses the located container number image,and uses the connected domain segmentation method to segment the characters.Then the characters are sent to the template matching algorithm for character recognition.Through theoretical analysis and experiments,the effectiveness of this method is proved,and compared with other methods,the recognition accuracy is significantly improved.
作者
姚砺
李莉莉
万燕
YAO Li;LI Lili;WAN Yan(School of Computer Science and Technology,Donghua University,Shanghai 201620,China)
出处
《智能计算机与应用》
2021年第7期95-101,共7页
Intelligent Computer and Applications