Contextual analysis of multisource Raster data by a region-based segmentation tool in a Voronoi structure

This paper presents a new concept of suitability assessment based on the development of a context-oriented analysis tool that handles multisource raster data considered as the multiple constraints encountered in a planning project. Region growing algorithm in a Voronoi neighbourhood structure is used to implement the developed concept. Two spatialization weighting models are considered. The first one uses a posteriori integration of spatial dimension where segmentation process is posterior to weighting process. The essential effect of this model is the regionalization of weights in order to reduce the amount of information to be processed in the subsequent steps to find solutions to spatial planning problems. In the a priori integration of spatial dimension, distinct regions are considered as the local context where weighting functions are applied. The developed tool has been experimented in the elaboration of suitability maps for power line implementation. Segmentation process is made inside a first-order Voronoi structure. Results show that a posteriori spatialization model consists on a redistribution of weights by the use of a regionalization tool delimiting homogeneous regions according to their weights. On the other hand, the a priori spatialization model presents the originality of adapting weighting functions to the local context. The use of semantic or statistic criteria in region growing process influence the partition scheme of the studied area and the subsequent processing. The strength of the developed tool lies in the huge possibilities of generating potential scenarios translating interactions between spatial features. is a reality. Most land management analysts limit their analysis to simplistic integration of information layers representing principal components of their projects. These methods based essentially on Boolean operators or algebraic functions lead generally to very restricted suitable areas especially if the constraint layers are numerous. In the present work, we explore new approaches for suitability mapping based on context explicit integration in the weighting process applied to a planning project. Raster maps segmentation tool is used in the developed approach. We first begin by a review of segmentation process, next we expose the suitability models development and the results of their application.