Label Peel-Off Inspection using Edge Extraction and Contour Selection.
Updated: May 9, 2021
Edge selection allows user to specify what kind of edges are looking for. In this example, we are using sub-pixel precise edges extraction to segment label objects from background, and then unites fragments lines to form a union collinear contour of the labels border. Typically, union collinear contour is part of the process of grouping and selecting extracted edges or lines to achieve meaningful entities.
When the labels border contour are formed, for those labels that consists of any peel off will eventually created an "open" contour where there are no joint from starting point to ending point. By exanimate these characteristic, we can say any contours that are not entirely closed will be a peel-off label.
To setup this inspection, simply add vision tools available in Squidify System:
Union collinear contour
Select "closed" contour to eliminate peel-off label(s)
Select correct contour perimeter to eliminate over or under size of the label(s)
Using edges extraction gives an advantages of objects at random position compared to specific ROIs examination. Moreover, Edges extraction use recursively implemented filters to extract sub-pixel precision of the objects gives higher accuracy and stability. Therefore, Squidify System offered simplicity implementation for these complex image processing technique at high performance operation.