| With the development of the Internet of Things technology,two-dimensional barcodes technology is becoming more and more mature,as an important source of information perception is widely used in production,manufacturing,logistics,warehousing,and other fields.QR code is the most widely used two-dimensional barcode because of its strong error correction ability and fast recognition in all direction.In the warehousing logistics and other scenarios,the use of QR code artificial landmarks as visual identification to study the localization technology of mobile robots,which avoids many factors and achieves the accurate localization of mobile robots by quickly and accurately identifying QR codes.However,in the actual location scenario,the images captured will be disturbed by various factors and resulting in the inability to extract feature information quickly and accurately,affecting the speed and accuracy of localization.In the visual localization of mobile robots based on QR codes,the key to achieve localization is how to locate accurately and identify the relevant information in QR codes.However,because the QR code image collected contains environmental background information,resulting in low QR recognition speed and recognition rate,so it is not possible to decode the QR code image directly.Besides,for the actual localization application scenario,it is not possible to determine in advance how many QR codes exist in the captured image,the simultaneous occurrence of multiple QR codes also presents a new challenge to the traditional QR code detection algorithms.Besides,the visual localization of mobile robots based on single QR code identification is intermittent,as well as the fixed driving path,and challenges in localization accuracy and reliability.This paper mainly focuses on the above-mentioned problems,the research contents and innovations are as follows:First,the research on QR code fast illumination equalization algorithm.In the complex lighting environment,there are many types of uneven lighting problems in the QR code images collected.Based on the symbolic characteristics of QR codes,a robust adaptive binarization algorithm is proposed.Experiments show that the algorithm can quickly eliminate the influence of uneven illumination of QR code images,and the algorithm has a better effect on different uneven illumination,which can meet the requirements of fast and high recognition rate in real-time scenes.Second,the research on multi-QR codes fast detection algorithm.In the actual localization application scenario,there is an unknown number of QR codes in the captured image,and the traditional QR code detection algorithms cannot quickly locate multi-QR codes.By combining the traditional QR code detection method with the network learning method,this paper proposes a multi-QR codes fast detection algorithm based on the characteristics of rich information and dense distribution of QR code edge.After experimental verification,the proposed multi-QR codes detection algorithm can achieve high-progress detection of a single picture containing 160 QR codes in 0.6 seconds.Finally,the research on multi-QR codes localization algorithm.There are intermittent problems in the visual localization of mobile robots based on individual QR code identification,and the accuracy and reliability of localization need to be improved.Based on the proposed multi-QR codes fast detection algorithm,this paper designs a localization scheme for mobile robots,which determines the direction and actual position of the mobile robot by detecting the location of multiple QR codes and the encoding information identified,and achieves higher precision and more reliable localization.Besides,through the construction of the experimental scene for localization testing,the high precision localization of the proposed multi-QR codes localization algorithm is verified,which can estimate the position of the mobile robot effectively.It provides more accurate localization information for autonomous navigation. |
PDF Full Text Request