Sub-pixel mapping of remotely sensed imagery is a post-processing method of pixel unmixing in order to solve the issue of spatial uncertainty of mixed pixels. With the fraction images derived from pixel unmixing technologies as input

the spatial distribution of each land cover class within mixed pixels is determined and a sub-pixel scale land cover map is generated according to the spatial correlation principle. This paper firstly introduces the concept of sub-pixel mapping and analyses the theoretial models and currently used algorithms. Then

the main sources of errors

different accuracy assessments of the model and the ways to present the uncertainty of the model are summarized. The main methods for increasing the accuracy of sub-pixel mapping by ancillary data are also discussed. Finally

this paper concludes with an outlook and several key techniques of sub-pixel mapping.