The work aims to quickly and accurately detect common printing defects in beer box. With beer box surface paper as the detection target, the common defect features in beer box printing were detected by the template matching method of constructing variation model by extracting the shape and gray value information from the template image. Printing quality was judged according to the defect inspection results, and the effect of the quality inspection scheme was evaluated through inspection and verification experiment. According to the test experiment and accounting of the 500 images collected, the average accuracy of this method was 96.18%, the rate of missed detection was less than 0.9%, the rate of misjudgment was 3.08%, and the average detection time was less than 10 ms. Detecting quality of offset printing products such as beer box surface paper with this method has excellent and stable results. It can accurately detect the defects with high precision requirement, such as small scratches, and the detection speed is faster than other methods.