Modélisation 3D des façades de bâtiments des anciennes Médina

The static terrestrial LIDAR is a tool of surveying which present a vital source of 3D data in the heritage buildings evaluation and monitoring. Through automated 3D modeling process, manipulating this data source in communication tools becomes more flexible. The 3D modeling process imposes on essential tasks including segmentation. Much of segmentation approaches focus on the use of geometric components by exploiting the mathematical representation of a plane or curved area. Due to the complexity of the facades architecture, segmentation using only the geometrical aspect is insufficient. Thus, a new segmentation approach is developed balancing the results of extracting homogeneous areas based on all colored point cloud components namely geometric data, the RGB values and intensities of facades details especially in old Medina building. The process consists of the planar segments extracting using the RANSAC algorithm. The result will be subject to radiometric segmentation through the introduction of color similarity criterion in the region growing algorithm and finally a fusion of classes based on the intensity’s similarity.