Research on the Barcode Deblurring Algorithm Based on a GAN

With the development of Internet of Things (IoT) technology, barcode automatic recognition systems play a crucial role. Traditional methods often perform poorly when processing blurred barcodes, which affects recognition and application performance. This paper proposes a barcode deblurring algorithm based on generative adversarial networks (GANs), aimed at overcoming the problem of insufficient barcode clarity in traditional image processing. First, the SE attention mechanism is combined with the aggregation residual block ResNeXt to form SE-ResNeXt, replacing the residual block ResNet, which accelerates the model’s convergence speed and enhances the stability of the training process. Second, the channel prior convolutional attention (CPCA) mechanism is introduced to improve the network's feature extraction ability and detection performance. The experimental results show that the proposed model achieves a peak signal-to-noise ratio (PSNR) of 30.48 dB, an improvement of 4.87 dB over the baseline network, and a structural similarity index (SSIM) of 0.9383, an improvement of 7.72%. The subjective visual deblurring effect is also promising, with restored barcode images showing clear edge contours and noticeable detail recovery.