Research on barcode image recognition of cigarettes

To enhance the efficiency and accuracy of cigarette barcode recognition within tobacco logistics systems, this study proposes a barcode recognition method that integrates multiple image preprocessing and geometric correction techniques. Initially, grayscale conversion and Gaussian filtering are employed for image denoising to improve the clarity of barcode images. Subsequently, the maximum between-class variance method is used for binary segmentation of the barcode region, while morphological processing techniques, such as the closing operation, are applied to connect discontinuous barcode regions for precise localization. To correct potential tilts in barcodes positioned on conveyor belts, bilinear interpolation is introduced for geometric correction, which not only maintains recognition accuracy but also improves computational efficiency. Experimental results indicate that this method achieves a recognition accuracy rate of up to 99.9% for barcodes with tilt angles less than 30°.