摘要 |
PROBLEM TO BE SOLVED: To efficiently detect a continuously present flaw part or the same flaw part. SOLUTION: A flaw detection method has a first stage for subjecting a repeating pattern P(n) [n is arrangement coordinate having an arrangement interval as a unit, and is 1, 2, 3...] and the adjacent pattern P(n+m) [m is 1 or -1] to difference processing to obtain difference data S(n), a second stage for setting the number K of patterns having possibility such that almost the same flaws continue, a third stage for detecting flaw-free patterns continuing over the number K from the obtained difference data S(n) and a fourth stage for specifying the flaw on the pattern P(n) based on the comparison of the difference data S(n), the pattern P(n) and the adjacent pattern P(n+m) based on the flaw-free patterns continuing over the number K in the difference data S(n). |