The work aims to improve the recognition rate of intelligent inspection of defective liner paper in multi-layer package and the rate of good product of package by optimizing the light source conditions in the intelligent packaging inspection system. In the intelligent packaging inspection system, the defective liner paper of different colors was detected by changing the wavelength of the light source, and the pixel number of defective liner paper area detection was analyzed to determine the appropriate light source wavelength. The chromaticity resulted in red, green and blue visible monochrome light which could not image the liner paper of the same color in the intelligent packaging inspection system, but could detect all the defective liner paper under white, infrared and ultraviolet light. Further analysis showed that the imaging effect of defective liner paper was the best when the intelligent packaging inspection system was under infrared illumination, and the recognition rates under white light, infrared light and ultraviolet light were 90.5%, 99.5% and 96.5%, respectively. The recognition rate under infrared illumination was the highest. In the actual production of the packaging workshop, in order to achieve the highest accuracy of the intelligent packaging inspection system, the light source can be chosen according to the chromaticity or the infrared light source can be directly chosen.