Title :
Category extraction using NOAA AVHRR images
Author :
Kawano, Koichi ; Kudoh, Jun-ichi
Author_Institution :
Graduate Sch. of Inf. Sci., Tohoku Univ., Sendai, Japan
Abstract :
We have developed a new image processing method for the analysis of NOAA AVHRR data using the 3 dimensional histogram (Kudoh and Noguchi 1991). This method is based on the 3 dimensional histogram technique that can visualize the same properties of multispectral images as a mass in the 3 dimensional space. This method is composed of 2 processes. One is the learning process that gathers 3 dimensional histograms of NOAA AVHRR image scenes classified as the sea category by oceanography researchers. The other is the recognition process that automatically extracts the sea category for the unknown image scenes by using the only one 3 dimensional histogram obtained from the learning process. A feature of this method is that the recognition rate of the sea category can be improved as much as you want by using the NOAA AVHRR data received every day. We applied two methods to extract the sea category of the NOAA AVHRR images and compared them with the results from oceanography researchers. The results showed a more than 90% recognition rate for almost all scenes
Keywords :
geophysical signal processing; image classification; oceanographic techniques; 3 dimensional histogram; NOAA AVHRR images; category extraction; image processing method; learning process; multispectral images; oceanography; recognition process; sea category; Brightness; Character generation; Clouds; Data mining; Histograms; Image processing; Infrared imaging; Layout; Multispectral imaging; Pixel;
Conference_Titel :
Geoscience and Remote Sensing Symposium, 2001. IGARSS '01. IEEE 2001 International
Conference_Location :
Sydney, NSW
Print_ISBN :
0-7803-7031-7
DOI :
10.1109/IGARSS.2001.977988
