A visual detection method for foreign body particles in bottled solution

Abstract

The invention discloses a visual detection method for foreign body particles in bottled solution, which comprises the following steps: S1, continuously collecting T frame sequence images of bottled solution movement; S2, performing sub-pixel precision registration on the acquired sequence images to obtain the registered sequence images; S3, extracting ROI effective detection region sub-images in each frame sequence image and performing image enhancement processing respectively to obtain enhanced images of each frame sequence image; 4, respectively calculatinga difference image between two adjacent enhanced image based on that enhanced images correspond to each frame of the sequence images, and performing gray threshold segmentation on each difference image to obtain a new image; S5, extracting contour regions in each new image, and discriminating whether the contour regions are foreign objects or not based on the size features of the contour regions; S6: Identifying whether the foreignobject target is a bubble interference. On the basis of recognizing the foreign object, the method can identify whether the foreign object is a bubble or not, which improves the reliability of foreign object detection results.

Description

technical field [0001] The invention belongs to the field of industrial detection and machine vision technology detection, and in particular relates to a visual detection method for foreign matter particles in bottled solution. Background technique [0002] It is easy to mix foreign matters in the bottled solution during the filling production process, especially for medical bottled liquid medicines, which may be mixed with foreign matters such as fibers, rubber crumbs, hair, and glass shavings during the filling production process. The difficulty of foreign matter detection is that there are many types of targets and insufficient texture features; there are many interference factors, suspended particles in the liquid medicine, interference from the bottle wall, air bubbles, ambient light, random noise, etc. will affect the detection results; The field requires high-speed online real-time detection of foreign particles. [0003] Existing foreign object detection methods mai...

AI Technical Summary

Problems solved by technology

Existing methods based on machine vision such as CN 104835166 A "Method for detecting foreign matter in liquid medicine bottles based on machine vision detection platform" are to preprocess the image and set ROI, and realize foreign matter detection through module matching and frame difference method, such as CN102519984A " Anti-vibration Machine Vision Detection Method for Foreign Objects in Liquid Medicine"The phase correlation method based on mountain-climbing search is used for matching preprocessing, but the genuine and defective products are only distinguished according to the area of ​​foreign objects; neither of the above two existing methods can eliminate the interference of air bubbles, As a result, the foreign object detection result may be air bubbles instead of real particle foreign objects, the accuracy is not enough, and the practicability is not strong

Images