c++ - Product inspection with Computer vision - edge detection with opencv -
i new user of opencv. doing project of performing product inspection opencv. planning extract edges of product , bad product compare edges maybe mean square difference. however, quite difficult extract edge first step.
good sample:good product [![enter image description here][1]][1] when use canny edge detection, edge of product (the blue part of picture) includes part of product , follows: edge of product [![enter image description here][2]][2]
i tried use adaptivethreshold make greyscale picture more clear , use edge detection. but, edge detected not expected because of many noise.
therefore, ask solution of extracting edge or better way of comparing product , bad product opencv. sorry bad english above.
this task can made simple if assumptions below valid. example:
- products move along same production line, , therefore lighting across product images stays same.
- the products lie on flat surface parallel focal plane of camera, rotation of objects around axis of lens.
- the lighting can controlled (this better edges, , used in production line image processing)
since cannot see images added, bit hard see situation is. if above assumptions valid, image differencing combined image signatures 1 possible approach.
another possibility train haar cascade classifier using , bad products. this, edges , taken care of. have collect lot of data , train classifier.
Comments
Post a Comment