Remote sensing meets psychology: a concept for operator performance assessment

Abstract

An often undervalued but inevitable component in remote sensing image analysis is human perception and interpretation. Human intervention is a requisite for visual image interpretation, where the interpreter actually performs the analysis. Although image processing became more and more automated, human screening and interpretation remained indispensable at certain stages. One particular stage where the operator plays a crucial role is in the development of reference maps. This is often done by a visual interpretation of an image by an operator. Although the result is crucial for adequately assessing automated systems' performance, the work of the human operator is rarely questioned. No variability is considered and the possibility of errors is not mentioned. This is an implicit assumption that operator performance approaches perfection and that infrequent errors are randomly distributed across time, operators and image types. Given that the existence of operator variability has been proven in several other related domains, for example, screening of medical images, this assumption may be questioned. This letter brings the issue to the attention of the remote sensing community and introduces a new concept quantifying operator variability. As the WAVARS project (web-based assessment of operator performance variability within remote sensing image interpretation tasks) will gain from a high amount of data, we kindly invite interested researchers to access the website http://wavars.ugent.be and take part in the test.