Method and system for detecting defects on surface of object

A system for detecting a defect on a surface of an object comprising:

• a platform for supporting the object;
• a plurality of illumination sources comprising at least one ambient illumination source and at least one dark field illumination source for illuminating the surface of the object;
• an imaging device for capturing images of the surface of the object under illumination conditions with the illumination sources;
• and
• an image processor for:
• processing the captured images with a plurality of image operations to detect an area of a potential defect at a location on the surface of the object;
• cutting the area of the potential defect from the processed images to sub images;
• stitching the sub images together to generate a hypothesis of the potential defect at the location on the surface of the object;
• classifying the hypothesis in the stitched image with a classifier to determine whether the potential defect is a true defect, wherein the classifier is trained with training data having characteristics of the true defect;
• and
• generating an output of the classification comprising the detected true defect and the location on the surface of the object.

A computer program stored in a non-transitory computer readable medium and executable in a computer for performing a method of detecting a defect on a surface of an object, wherein the computer stores images of the surface of the object under illumination conditions with illumination sources comprising at least one ambient illumination source and at least one dark field illumination source, wherein the method comprises steps of:

• processing the captured images with a plurality of image operations to detect an area of a potential defect at a location on the surface of the object;
• cutting the area of the potential defect from the processed images to sub images;
• stitching the sub images together to generate a hypothesis of the potential defect at the location on the surface of the object;
• classifying the hypothesis in the stitched image with a classifier to determine whether the potential defect is a true defect, wherein the classifier is trained with training data having characteristics of the true defect;
• and
• generating an output of the classification comprising the detected true defect and the location on the surface of the object.