In order to quickly and accurately obtain the position offset and inclination information of the paper-plastic composite bag during the correction process, this paper proposes a paper-plastic composite bag visual positioning algorithm based on improved Hough transform. First, the paper-plastic composite bag image on the platform was segmented by the Otsu method to remove the belt part; then the paper-plastic composite bag image was denoised by the variance-based differential filtering method to highlight the edge of the composite bag image; finally, the Canny calculation was used to obtain the edge points of the paper-plastic composite bag, and the improved Hough transform algorithm was used to extract the four edges of the paper-plastic composite bag as a straight line to solve the center position and inclination of the paper-plastic composite bag. The improved algorithm was superior to the traditional Hough transform algorithm in terms of line extraction accuracy and time-consuming. It only took 0.335 s to obtain the position and inclination of the paper-plastic composite bag, which reduced the time consumption by 76%. Therefore, the time consumption of the improved algorithm meets the requirements of deviation correction and location.