Abstract
Abstract. This paper describes a preliminary study on the image orientation acquired by a hyperspectral frame camera for applications in small tropical forest areas with dense vegetation. Since access to the interior of forests is complicated and Ground Control Points (GCPs) are not available, this study conducts an assessment of the altimetry accuracy provided by control targets installed on one border of an image block, simulating it outside a forest. A lightweight Unmanned Aerial Vehicle (UAV) was equipped with a hyperspectral camera and a dual-frequency GNSS receiver to collect images at two flying strips covering a vegetation area. The assessment experiments were based on Bundle Block Adjustment (BBA) with images of two spectral bands (from two sensors) using several weighted constraints in the camera position. Trials with GCPs (presignalized targets) positioned only on one side of the image block were compared with trials using GCPs in the corners. Analyses were performed on altimetry discrepancies obtained from altimetry checkpoints. The results showed a discrepancy in Z coordinate of approximately 40 cm using the proposed technique, which is sufficient for applications in forests.
Highlights
The monitoring of recovered and native forests is a widely recognized global need which requires updated geospatial information
Considering the band selected in the sensor 1 (689.56 nm), Figure 6 presents a graph generated with the root mean square error (RMSE) of Ground Control Points (GCPs) resulting from the Bundle Block Adjustment (BBA) for different weighted constraints in the camera Perspective Centre (PC) position
The results indicated larger RMSEs when GCPs were only used in one side of the image block, presenting values of 0.460-0.053 m in X and 0.035-0.046 m in Y
Summary
The monitoring of recovered and native forests is a widely recognized global need which requires updated geospatial information. In remaining forests, for example, the use of images collected by Unmanned Aerial Vehicles (UAVs) is feasible due to the lower costs and the possibility of images acquisition with suitable spatial and temporal frequency. New types of hyperspectral sensors have been introduced to UAV applications, such as the Rikola camera with a Fabri-Perot interferometer (FPI), which acquires sequence 2D images in frame format. A review presented by Aasen et al (2015) reported important studies and the potentiality of hyperspectral sensors to derive information, e.g., about vegetation, plant diseases, environmental conditions, and forest. Honkavaara et al (2013) performed experiments using UAV with a FPI-based hyperspectral camera to collect hyperspectral and structural information and to estimate plant height and biomass. The trials demonstrated great potential for precision agriculture and indicated feasibility for other research topics
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
