Sie befinden Sich nicht im Netzwerk der Universität Paderborn. Der Zugriff auf elektronische Ressourcen ist gegebenenfalls nur via VPN oder Shibboleth (DFN-AAI) möglich. mehr Informationen...
2D barcodes are widely used in many commercial applications where a scanning device is normally used to capture them. When mobile phones are used to capture 2D barcodes, the obtained images are usually distorted due to cheap camera lens and sensors, handshake and poor lighting conditions. These badly distorted images require a long decoding process which results in an error message generated or wrongly decoded information. In this paper, we propose a two-stage quality measure for the mobile phone captured 2D barcodes in order to reject those poor quality images. The proposed method is based on the global bimodal distribution features and the local finder pattern detection. Experimental results on QR code images show that the proposed two-stage quality measure has 97.64% prediction accuracy with an average run time of 110ms by rejecting distorted undecodable barcode images in advance. The proposed method also has good generalizability to “unseen” camera models and performs well under different lighting conditions. Experiments on data matrix images show that our quality measure can be extended to 2D barcode patterns with similar features.
► A quality measure for mobile phone captured 2D barcode images is proposed. ► Two sets of global and location complementary features are extracted. ► Classification accuracy over 97% is achieved on a large QR code images database. ► The measure has good generalizability to different camera models, lightings. ► The features can be adapted to other 2D barcode patterns with similar structure.