Algorithm Research on Medicinal Glass Bottles Defect Detection of Printing Words
In the process of transmission,medicinal glass bottle may occur translation,rotation or affected by noise.At present,printed words of bottle have no effective way to detect.In order to accurately detect the defects of printed words,two matching methods are compared,Harris combined with normalized cross correlation method (NCC) and SIFT combined with Euclidean distance method.Two methods are applied to the the defect detection system,the most effective method is chosen.The experiment results show that compared with Harris,SIFT method can more accurately extract the image feature points,and the correct matching rate with Euclidean distance is high,which can achieve the desired experimental result.
SIFT algorithmHarris algorithmfeature point extractionimage registrationmedicinal glass bottles printing word detection
NETL
NSTL
万方数据
维普
