摘要 |
<P>PROBLEM TO BE SOLVED: To prevent reduction of a classification correct answer rate of defective images caused by a difference in nature between images due to a difference between observation devices when images captured by a plurality of different observation devices are mixed to be input in a device for automatically classifying images imaging defects on a semiconductor wafer for each defined class by a user. <P>SOLUTION: In an automatic image classification device to which defective images captured by a plurality of observation devices are input, when a recipe is created, an image processing parameter is adjusted and a classification identification surface is created for each observation device. During image classification, the observation device that has captured a defective image is identified based on incidental image information. Image processing and classification processing are carried out by using the image processing parameter and the classification identification surface according to the observation device that has captured the image. To efficiently adjust an image processing parameter for each observation device, an appropriate image processing parameter is automatically adjusted based on an instructed defective area. <P>COPYRIGHT: (C)2012,JPO&INPIT |