摘要 |
PROBLEM TO BE SOLVED: To provide a method for generating a data map for determining the surface state of an object, having a plurality of continuous constituting units of similar textures. SOLUTION: This method includes a step of preparing a plurality of samples of an inspection domain, including a domain required to be determined, a step of extracting a characteristic vector from the sample, and a step for inputting the characteristic vector into a self-organized map and making it to learn. The method also includes a step for forming a cluster by grouping adjacent neurons corresponding to each of the same sample in the learned self-organized map, and a step for classifying each cluster into one corresponding to a prescribed surface state having flaws, nicks, exfoliation or the like and a different one, and generating the determination data map. Since the data map for automatically determining the surface state of the object having the plurality of continuous constitutional units of the similar texture is generated by utilizing self-organization learning of a neural network, the determination data map having high reliability can be acquired. COPYRIGHT: (C)2006,JPO&NCIPI
|