Abstract
This paper makes contribution to the problem of automatic detection of GCP's for AVHRR imagery. The approach presented in this paper, consider improvement to edge detection accuracy for AVHRR imagery using an NDVI threshold, adaptive thresholding and spatial filtering techniques. Two different matching algorithms based on the edge extraction approach were applied to the automated detection of GCP's for the complete geometric registration of AVHRR data. The first is a line matching algorithm for the AVHRR image in satellite projection and the second is a pattern matching algorithm for the ground projected image. In order to apply this method to operational processing of AVHRR data, coastline vector data, GCP and GCMP data sets and elevation data were created. The importance and novelty of the suggested approach is that GCP identification and geographical data set creation have been integrated and an algorithm allowing fully automatic GCP detection for AVHRR was developed. The method has been applied to geometric registration of 10 day composite AVHRR data for the Asian region. In the future, it can be applied to operational processing of AVHRR data on a global scale.
