Sequential approach for automatic defect recognition

A method of automatic defect recognition comprising:

• receiving a initial set of inspection image data of a scanned object from a scanning machine;
• specifying at least one of:
• sensitivity and specificity associated with the defect recognition;
• applying a first image analysis algorithm to the initial set of inspection image data;
• based on the applying of the first image analysis algorithm, removing from the initial set of inspection image data a first set of defect-free image regions in accordance with the at least one of:
• sensitivity and specificity, further reducing number of voxels to be processed in subsequent image analysis and thereby retaining a first set of analyzed inspection image data;
• applying a second image analysis algorithm to the first set of analyzed inspection image data, wherein the second image analysis algorithm has a higher computational cost than the first image analysis algorithm;
• and
• based on the applying of the second image analysis algorithm, removing from the first set of inspection image data a second set of defect-free image regions in accordance with the at least one of:
• sensitivity and specificity, further reducing number of voxels to be processed in a subsequent image analysis and thereby retaining a second set of twice-analyzed inspection image data;
• wherein the first and the second image analysis algorithm comprise a sequential and cascaded approach.