1 |
ELIMINATION OF LEAF ANGLE IMPACTS ON PLANT REFLECTANCE SPECTRA BASED ON FUSION OF HYPERSPECTRAL IMAGES AND 3D POINT CLOUDSLibo Zhang (13956072) 13 October 2022 (has links)
<p>In recent years, hyperspectral imaging technologies have been broadly applied to evaluate complex plant physiological features such as leaf moisture content, nutrient level and disease stress. A critical component of this technique is white referencing used to remove the effect of non-uniform lighting intensity in different wavelengths on raw hyperspectral images. Based on the literature, the leaf geometry (e.g., tilt angles) and its interaction with the illumination severely impact the plant reflectance spectra and vegetation indices such as the normalized difference vegetation index (NDVI). This thesis is aimed to address the issues caused by the tilt angles across the leaf surface. To achieve this, two methods based on the fusion of the hyperspectral images and 3D point clouds were proposed. The first method was to build a 3D white reference library in which a point with almost the same tilt angle, height and position with the pixel on the plant leaf can be found, and then the white reference spectrum at that point can be used to calibrate the raw spectrum of the leaf pixel. The second method was to observe and summarize how the plant spectra and NDVI values changed with the leaf angles. Using the changing trends, the original NDVI and spectra of leaf pixels at different angles can be calibrate to a same standard as if the leaf was imaged at a flat and horizontal surface. The approach was called 3D calibration. The results showed that the NDVI values significantly changed with leaf angles and the changing trends differed between the corn and soybean species. To evaluate the performance of 3D calibration, 180 soybean plants with different genotypes, nitrogen (N), phosphorus (P) and water treatments were grown in the greenhouse. Each plant was imaged in three systems: the high-throughput greenhouse hyperspectral imaging system, the indoor desktop imaging system with a visible-near infrared (VINIR) hyperspectral camera and an Intel RealSense depth camera and the handheld device hyperspectral imaging system. In the greenhouse system, the whole canopy was captured. In the indoor desktop system, the partial canopy was captured because of the space limitation and the top-matured leaf (the middle leaf of the uppermost matured trifoliate) was focused. The proposed 3D calibration was applied on the top-matured leaf to remove angle impacts. In the handheld device system, the flat top-matured leaf was captured. After done with imaging work, the plants were harvested to collect the ground truth data such as relative water content (RWC), N content and P content. Combined with the ground truth data, the NDVI values from three systems were used to discriminate different genotypes and biochemical treatments, whereas, the spectra from three systems were used to build partial least squares regression (PLSR) models for N, P and RWC. The results showed that the averaged tilt angles of top-matured leaves were impacted by different treatments. For instance, the low-nitrogen (LN) plants showed significantly higher leaf angles than high-nitrogen (HN) plants; the leaf angles on water-stressed (WS) plants were higher than those on well-watered (WW) plants. The leaf angles carried some signals that influenced not only the NDVI discrimination but also the PLSR modelling results. The signals were lost after 3D calibration. For the top-matured leaves, the discrimination and modelling results after 3D calibration in the indoor desktop system were close to those from the flat leaves in the handheld device system. The proposed 3D calibration approach has a potential to eliminate leaf angle impacts.</p>
|
Page generated in 0.1197 seconds