Generation of multispectral point clouds by a mobile terrestrial system

Abstract

Mobile Mapping Systems (SMM), combining technologies of geodetic positioning (navigation) and remote sensors, have been used to monitor agricultural crops, as they allow dynamic and systematic data acquisition at a low cost. The acquisition of images and navigation data, combined with processing with photogrammetric techniques, makes it possible to generate detailed digital models of plants and soil with high spatial and temporal resolution. These digital models can be used for complex analyzes of the entire structure of the cultivated area. This study aims at generating multispectral point clouds using commercial software tools such as Metashape and lower-cost sensors such as Agrowing. These cameras use commercial digital cameras to which special objectives with multiple lenses and filters are attached. They have several advantages compared to other multi-band sensors: lower price, lightweight and use the structure of proven commercial cameras, no synchronization issues and provide high-resolution images. By using the structure of consumer cameras, the integration with other devices, such as navigation systems, is straightforward. In this work, two 6-head Agrowing cameras will be used with a dual-converged assembly, which allows a vertical coverage of 70º, to simulate mobile mapping systems. This system will be manually shifted on short bases (about 1 m) in order to produce a dense cover over citrus plants. Control points and exposure stations will be surveyed by GNSS. These images will be oriented in the Metashape software, and the cameras will be previously calibrated in a terrestrial calibration field. Dense clouds will be generated for each of the 14 spectral bands and converted to a voxel structure.