Answers are grounded in this paper and cite their sources.
We got some kirids ofcbservation data by artificial satellites. In recent years. It was thought that the accuracy of classification should rise when the rumber of bands to analyze increase. Bit if unfitted bands were used, The accuracy should be down. The purpose of this paper are two. The first one is to analyze the appropriate number of bands to use. The second one is to define the choice index of bands to use. It was concluded that the accuracy increased with the number of bands increasd until 6 bands. But in the case to use 10 bands, the accuracy fell. About analysis time, it increased according to the number of bands. The choice index (the expectation correct answer percentage) was defined to search fitted band in many bands. Expectation conect answer percentage is calculated using the standard regular accumulate distribution function from the distance of maharanobis distance during eachcategory of dassifcation. To use the index, the appropriate combination of bands is found with comparatively simpoe calculation.
No takes yet. Share an insight, caveat, or question.
Naoki TAKAGI (1997) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: