It is a widely used approach to perform band combinations in analysis of remotely sensed data. However

how to select best bands and how to determine best band combinations

are two key problems in band combination analysis which have not solved yet. One of major reasons is lack of deep understanding to band combination mechanism. Concepts of projective line and isopleth are used in this paper to analyze the geometric characteristics of each band arithmetic combination （such as addition

subtraction

multiplication

division and square sum etc.） in two dimension plane space so as to reveal intrinsic mechanism of band combination. Provided a little amount of ground feature data are available

band conbination scheme can be determined according to data distribution characteristics （i.e. positive or negative distribution）. Also

best bands can be determined depending upon whether the isopleth could most efficiently distinguish characteristic objects from non-characteristic objects background or not. Then

best band combinations can be determined according to whether projective area of sopleth could most efficiently enrance characteristic object information and suppress non-characteristic object information or not. It is proved through the case study by analysis of Thematic Mapper CCT data in Inner Mongolia

（China） that the approach discussed in this paper is practical and efficient. As to combination mechanism of more than two bands

further analysis is expected.