Human Centric Approach to Inhomogenious Geospatial Data Fusion and Actualization

This paper describes a research which attempts to combine the advantages of human analysts and computer automated processing for efficient human computer symbiosis in geospatial data fusion. Specifically, experiments performed were related to the analysis of the potential use of inhomogeneous (composed of different sources) stereo pairs for mapping dataset actualization. Inhomogeneous stereo pairs were combined with images of the map to be updated along with actual aerial images of the same territory. The anaglyphic product obtained after image processing of such stereo pairs was demonstrated to human analysts (subjects) and stereo perception of such stereo pairs was achieved. The most interesting finding of this experiment is the fact that some objects existing only on the aerial photo appeared in the inhomogeneous stereo pairs as 3D. This effect is caused by phenomena within the human eye-brain system known as human stereopsis which is widely deployed in photogrammetry. For the quantities measurements of the effect obtained, an eye-tracking system was deployed. Analysis of human eyemovements (driven by conscious and subconscious brain processes) while perceiving an inhomogeneous stereo dataset -provides a unique opportunity for the human computer symbiosed geospatial systems. There are two potential outcomes of such approach: a) interpretative – analysts’ gaze-fixation zones can help to localize the areas where mapping dataset should be updated b) quantitative – processing of eye fixations geometry during stereo model perception allows to transform the virtual 3D model to a geometrical one based on binocular summation measurements deploying eye-tracking.