The work aims to accurately locate the outer edge contour of the existing plastic bag container, so as to ensure that the outer edge contour of the transparent packaging container can be applied in daily production. According to the image information collection, recognition algorithm, Sobel algorithm, position calibration, SVM results comparison and other parts, the machine vision detection system for plastic bag container determined the key points of the transparent packaging edge calibration algorithm. Based on the traditional frequency domain transformation, Sobel algorithm was improved for analysis, to effectively improve the recognition rate for the outer edge contour of the plastic bag container, and clarify the detection points of this kind of packaging. With the actual liquid transparent medical packaging bag as an example, the feature extraction method of the outer edge of the packaging was demonstrated, and the vision characteristics of the packaging edge were strengthened. Finally, the SVM results were compared with the experimental results. This vision method realized a calibration accuracy of 94% for the plastic bag container by the machine vision and was applicable to 10% of edge calibration accuracy of manipulator. Based on the Sobel algorithm and optimization measures, the calibration method of plastic bag container by machine vision is determined, which meets the requirements of effective positioning of plastic bag container production.