| 摘要 |
<P>PROBLEM TO BE SOLVED: To enhance precision in detecting an abnormal shade candidate, and to change the detection condition of the abnormal shade candidate. <P>SOLUTION: In an SVM (support vector machine) organization processing, teacher data to be a support vector is obtained in the respective categories of the abnormal shade and a normal shade, based on a teacher data group. Then, SVM is organized by calculating an identification function by using a maximum margin area, based on the teacher data being the support vector. In this case, it is determined whether or not the teacher data defined as SV is appropriate. When it is determined that the data is inappropriate, the teacher data defined as the inappropriate SV is deleted, so as to calculate the SV and the identification function again. <P>COPYRIGHT: (C)2005,JPO&NCIPI |
