目的 针对传统喷码检测方法计算量大、字符区域定位不显著、识别准确率较低等不足，提出一种基于机器视觉的柔性包装袋喷码缺陷检测方法。方法 以柔性包装袋上喷码图像为研究对象，以滤波抑噪、阈值处理等技术对图像进行预处理，运用YOLO-V3网络模型对字符区域进行定位，并采用阈值和非极大值抑制算法提高喷码区域定位的显著性，通过改进AlexNet网络结构、运用多特征融合运算等方法，获取更为丰富的图像卷积特征，实现字符串的整体识别，从而提高喷码缺陷识别的准确率。结果 将YOLO-V3联合改进AlexNet的检测方法与传统喷码检测方法进行对比，结果表明，所设计喷码缺陷检测方法的分类准确率达到99.39%。结论 基于机器视觉的柔性包装袋喷码缺陷检测方法在模型计算量、字符区域定位显著性和字符识别准确率都有一定的优势，并有效解决了字符串整体识别的问题。

Aiming at the disadvantages of traditional coding detection methods such as large amount of calculation, insignificant character area positioning, and low recognition accuracy, a machine vision-based flexible packaging bag coding defect detection method is proposed. With the coded image on the flexible packaging bag as the research object, the image is preprocessed by filtering and noise suppression, threshold processing and other technologies. YOLO-V3 network model is used to locate character area, and threshold and non-maximum suppression algorithm is used to improve the significance of coding area positioning. By improving the AlexNet network structure and methods such as multi-feature fusion operations, more abundant image convolution features are obtained and the accuracy of coding defect recognition is improved. By comparing the detection method of YOLO-V3 jointly-improved AlexNet with the traditional coding detection method, it is found that the classification accuracy of the designed defect detection method is 99.39%. The machine vision-based flexible packaging bag coding defect detection method has certain advantages in the amount of model calculation, the significance of character area positioning, and the accuracy of character recognition, and it can also effectively solve the problem of string overall recognition.