抄録
Time series analysis is one of the important methods for understanding the change of a phenomenon. But, each image has a different distortion generally. Therefore, geometric correction of images is needed. Up to now, many discussions about the accuracy of geometric correction have been held and many geometric correction algorithms about NOAA/AVHRR images have been proposed. However, most of them use GCP (ground control point) where a user is required to specify GCP manually. In this research, special attention is given to this point and automatic GCP creation method is proposed. Required inputs of this method are only an AVHRR image and a standard map. First, a binary image which expresses land and a water area is created by calculating NDVI of an AVHRR image. Next, the boundary of land and a water area is detected from the binary image, and reference windows are created from a digital map. Finally, the boundary of AVHRR image and the reference windows are compared and the best match position is determined as GCP. No manual process is needed by this proposed method. Therefore, processing time is shortened and a user's labor is mitigated. The result of the geometric correction algorithm using this proposed method shows the mean errors are less than 1 pixel.
