Earth observation big data analytics in operating mode for GIScience applications. The (GE)OBIA acronym(s) reconsidered - Université de La Réunion Access content directly
Conference Papers Year : 2018

Earth observation big data analytics in operating mode for GIScience applications. The (GE)OBIA acronym(s) reconsidered


In 2006, to bridge the gap between geographic information systems and remote sensing (RS), terms object-based image analysis (OBIA) and geographic OBIA (GEOBIA) were introduced by geographic information science (GIScience) as viable alternatives to pixel-based image analysis. The eCognition commercial software for image analysis, brought to market in 2000, fostered the emergence of a GEOBIA community. Adopted as the GEOBIA de-facto reference standard, the inherent potential of eCognition has often been reduced to a heuristic-based, semi-automatic and site-specific workflow, starting from its multi-resolution image segmentation first stage. To date the GEOBIA community has grown somewhat self-contained, little aware of principles on human vision and human visual perception proposed by neural science, psychophysics and computer vision (CV) in the multi-disciplinary realm of cognitive science. Well-grounded principles of vision in cognitive science suggest the GEOBIA community to reconsider acronyms OBIA/GEOBIA, as recommended below. Vision, encompassing both human vision and CV, is an inherently ill-posed cognitive (information-as-data-interpretation) process whose goal is scene-from-image reconstruction and understanding. In vision, spatial information typically dominates color information, e.g., human panchromatic and chromatic vision are nearly as effective. Vision is ill-posed because affected by: (I) data dimensionality reduction from the 4D space-time scene-domain to the (2D) image-domain, and (II) the semantic information gap from ever-varying sensory data to stable percepts in the mental world (conceptual world model). Ill-posed vision is difficult to solve: it is non-polynomial-hard in computational complexity and requires a priori knowledge in addition to sensory data to become better posed for numerical solution. Stemming from these principles the first conclusion is that acronym OBIA is vague because word object' may refer to either a sub-symbolic planar object in the image-domain or a real-world symbolic object in the 4D spatio-temporal scene-domain. Synonym of segment-based image-analysis? predated in the RS and CV literature, acronym OBIA should be reformulated as image-object-based image analysis. The second consideration is that the OBIA paradigm lacks solutions to better constrain an inherently ill-posed image-object-based interpretation second stage, let free to be implemented as either: a 1D image analysis approach, equivalent to 1D orderless pooling of spatial context-sensitive image-objects, where spatial topological information is totally lost in contrast with the OBIA principle that visual spatial information dominates color information, or a 2D topology-preserving image analysis approach, where primary spatial topological/non-topological information components are fully exploited together with secondary color information. The third consideration is that acronym GEOBIA, originally introduced to emphasize Earth image understanding, is misleading. If it reads as image analysis based on geographic image-objects, then it is a contradiction in terms because no image-object in the image-domain features geographic coordinates pertaining to the scene-domain. If it reads as image analysis based on geographic objects in the scene-domain, then it is questionable, because vision means scene-from-image reconstruction and understanding and not vice versa (actually, psychophysics proved that vision combines inductive with deductive inference). Following these observations, we propose acronym "EO4GIScience2D" to mean "EO big data analytics in operating mode for GIScience applications, constrained by 2D image analysis in cognitive science."
Fichier principal
Vignette du fichier
GEOBIA20018_AboutTheGeobiaAcronym_FullPaper.pdf (1.21 Mo) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-01958406 , version 1 (12-04-2019)


  • HAL Id : hal-01958406 , version 1


Andrea Baraldi, Stefan Lang, Dirk Tiede, Thomas Blaschke. Earth observation big data analytics in operating mode for GIScience applications. The (GE)OBIA acronym(s) reconsidered. GEOBIA 2018 - From pixels to ecosystems and global sustainability ​, Centre d'Etudes Spatiales de la BIOsphère (CESBIO); Office national d'études et de recherches aérospatiales (ONERA); Espace pour le développement (ESPACE DEV); UMR TETIS, Jun 2018, Montpellier, France. ⟨hal-01958406⟩
181 View
147 Download


Gmail Mastodon Facebook X LinkedIn More