Abstract
For oceanography researchers, it is very important to use satellite images for their research. But they had to classify the sea category from unknown images before using them in the practical work. So far we proposed the three-dimensional histogram method to automatically classify the sea category from NOAA AVHRR images to help classifications for oceanography researchers. But using this traditional method, errors were present between the histograms of the sea and the other when they classified many images. These errors affect the extraction of the sea because these errors are stored in the histogram of the sea. It is a serious problem when we add the results of daily analysis to the histogram. In this paper, we proposed a new method based on the three-dimensional histogram to improve this problem. It uses frequencies of these histograms to decide the boundary of the sea category and others. It is possible to decrease errors, and it is useful for long-term analysis. The presented method was successfully applied to channel-1, -2, and -3 of NOAA AVHRR images in 1990, and it succeeded to achieve the sea detection rate of 94% on average.
| Original language | English |
|---|---|
| Pages | 3462-3464 |
| Number of pages | 3 |
| Publication status | Published - 2003 Nov 24 |
| Event | 2003 IGARSS: Learning From Earth's Shapes and Colours - Toulouse, France Duration: 2003 Jul 21 → 2003 Jul 25 |
Other
| Other | 2003 IGARSS: Learning From Earth's Shapes and Colours |
|---|---|
| Country/Territory | France |
| City | Toulouse |
| Period | 03/7/21 → 03/7/25 |
Keywords
- NOAA AVHRR
- Sea Detection
- Three-Dimentional Histogram
ASJC Scopus subject areas
- Computer Science Applications
- Earth and Planetary Sciences(all)