The invention relates to an epitaxial
wafer defect positioning method based on
machine vision, and relates to the field of defect localization,
dislocation and crack scale parameters are predicted by collecting a control parameter sequence in the epitaxial
wafer growth process, then the number of
microscopic image collection areas is configured, a recognition area is selected on an epitaxial
wafer, and microscopic images are collected;
dislocation and crack identification is carried out by utilizing
machine vision, the propagation direction of the
dislocation is predicted when the dislocation is identified, meanwhile, the crack
influence coefficient when the dislocation and the crack coexist is processed to correct the number of crack areas, and finally, identification and correction are continued according to the dislocation propagation direction. The technical problem that the defect detection efficiency and accuracy are insufficient due to the fact that
control parameters, defect propagation rules and mutual influences in the epitaxial wafer growth process are difficult to comprehensively consider is solved, accurate positioning of epitaxial wafer defects is achieved, and the defect detection accuracy is improved. And the efficiency and the accuracy of epitaxial wafer defect detection are obviously improved.