CodePudding user response:
Binarization threshold segmentation to extract the ROI area open operation to remove small interference at the same time makes the crack expand facilitate testing with no crack bottle image do subtraction to find differences is crack, or the normal bottle image modeling, through the model to match all kinds of bottle image and find out the defect,CodePudding user response:
1/create a vial of the template (according to your image NCC/shape can be), to create the relative position of interested area (do you want to test part)2/positioning vial, affine detection area, narrow domain
3/the following algorithm is many, said also, on the second floor with edge can also be (find grayscale change obvious point coordinates)
CodePudding user response:
With threshold value method to determine the position of the bottle, fitted the coordinates of the rectangle to draw bottle as a benchmark, then according to the relative distance of the bottom of the bottle to bottle, for you to test the bottle to draw ROI area, by using the threshold value method or the implementation of edge detection to extract crack and area detection, boy!CodePudding user response:
Such test is not much use, you have to take detection of liquids, is the significance to practical production, that the only filling after many working procedure, even may lead to crack of the bottle, so you this test, it is best before packaging