Search

To perform classification on remotely sensed imagery data, a small number of pixels or fields need to be labelled. It has been found that this labelling task requires many auxiliary materials such as crop calendar, soil data, cropping practice and a weather summary. This task is at the present time manually performed by photo-interpreters. It is time-consuming and liable to human error. The paper describes a computer based interactive color display system for assisting photointerpreters. Its objective is to reduce contact time and to increase labelling accuracy. This system has been designed to extract features from a temporal series of MSS data and to display these features using color graphic techniques. Special attention was paid to convert crop calendar information to quantitative information consistent with observed data. More specifically, descriptive crop phenology on a crop calendar was converted into quantitative growth index curves, to which observed features derived from MSS data are directly comparable. This system has been designed for feasibility demonstration only and although it is still in the evolutionary stage, it has already demonstrated that the method to be presented offers an effective solution to alleviating many problems associated with the current manual labelling process.