特征空间注意力微小二维码重构算法

Micro QR Code Reconstruction Algorithm Based on Feature Space Attention

针对边长为5 mm以下的微小二维码直接识别正确率低,且通过超分辨率技术放大后的二维码会出现不规则形变和噪声干扰等问题,提出一种特征空间注意力微小二维码重构算法。该方法构建环形特征空间注意力权重滤波器模板,通过该权重滤波器计算图像块的加权平均灰度值,利用图像块的最佳分割阈值进行阈值分割,完成二维码高精度重构任务。实验结果表明,通过特征空间注意力方法重构出的二维码较其他二维码处理方法识别准确率显著提升。

Aiming at the low direct recognition accuracy of the micro QR code with side length less than 5 mm,and the irregular deformation and noise interference problems of the micro QR code after amplified by super resolution technology,this paper proposes a micro QR code reconstruction algorithm based on feature space attention.In this method,the attention weight filter template of the ring feature space was constructed,and the weighted average gray value of the image block was calculated by the weight filter.Then,the optimal segmentation threshold of the image block was used to perform the threshold segmentation,and the high-precision reconstruction of the QR code was completed.The experimental results show that compared with other QR code processing methods,the recognition accuracy of QR code reconstructed by the feature space attention method is significantly improved.