The work aims to effectively filter out the image noise in the visual inspection system of bubble cap drug package, to ensure that such tasks as image segmentation and feature extraction of subsequent drugs with bubble cap package are smoothly carried out. The mathematic model of guidance filtering was analyzed and improved based on such analysis. A new adaptive Gaussian-guided image filtering algorithm for bubble cap drug package was proposed. A more rational parameter optimization algorithm was adopted and the Gauss window was used to replace the traditional box window. The simulation results showed that the adaptive Gauss-guided filtering algorithm could effectively remove the noise in the image, while preserving the edge and details, etc. of the original image to the maximum extent. The proposed algorithm has certain guidance significance for the image processing and analysis of bubble cap drug package at later stage.