With the development of science and technology

new remote sensing devices are lunched on schedule. Optical satellite data have been undergone a long time renewal and accumulation. How to reveal hidden information from huge amount of remote sensing data and let it knowledgeable is new challenge for us. In this article

first we analysis the relation between material

structure spectral and remote sensing information

formulate space angel model. Then based on pixel vector and band vector tendency face of multi-band remote sensing information

we analysis the results of“not proper”in the cases of mining some information from remote sensing data when using the software algorithms which are selected according to“the bast optimum”of the statistics. Finally

we illustrate the mathematical principles and effectiveness of algorithms that we used in mining multi-band TM data.