Three-Dimensional Point Cloud Reconstruction and Morphology Measurement Method for Greenhouse Plants Based on the Kinect Sensor Self-Calibration

Abstract

Plant morphological data are an important basis for precision agriculture and plant phenomics. The three-dimensional (3D) geometric shape of plants is complex, and the 3D morphology of a plant changes relatively significantly during the full growth cycle. In order to make high-throughput measurements of the 3D morphological data of greenhouse plants, it is necessary to frequently adjust the relative position between the sensor and the plant. Therefore, it is necessary to frequently adjust the Kinect sensor position and consequently recalibrate the Kinect sensor during the full growth cycle of the plant, which significantly increases the tedium of the multiview 3D point cloud reconstruction process. A high-throughput 3D rapid greenhouse plant point cloud reconstruction method based on autonomous Kinect v2 sensor position calibration is proposed for 3D phenotyping greenhouse plants. Two red–green–blue–depth (RGB-D) images of the turntable surface are acquired by the Kinect v2 sensor. The central point and normal vector of the axis of rotation of the turntable are calculated automatically. The coordinate systems of RGB-D images captured at various view angles are unified based on the central point and normal vector of the axis of the turntable to achieve coarse registration. Then, the iterative closest point algorithm is used to perform multiview point cloud precise registration, thereby achieving rapid 3D point cloud reconstruction of the greenhouse plant. The greenhouse tomato plants were selected as measurement objects in this study. Research results show that the proposed 3D point cloud reconstruction method was highly accurate and stable in performance, and can be used to reconstruct 3D point clouds for high-throughput plant phenotyping analysis and to extract the morphological parameters of plants.

Combination of an Automated 3D Field Phenotyping Workflow and Predictive Modelling for High-Throughput and Non-Invasive Phenotyping of Grape Bunches

“…For one-year-old poplar plants, 90 • front view scans showed the highest similarities in comparison to 360 • scans for the estimation in total biomass and height. Sun and Wang [55] used a Kinect v2 sensor to scan tomato plants in the greenhouse. They used different recording angles (three vs. four recording angles) and converted the corresponding RGB-D images into 3D point clouds.…”

A Low-Cost 3D Phenotype Measurement Method of Leafy Vegetables Using Video Recordings from Smartphones

“…Yang et al [14] proposed a 3D shape measurement method for fruit tree canopies. Sun et al [15] proposed a Kinect-based 3D phenotype calculation method for greenhouse plants. Wang et al [16] evaluated the phenotypes of corn at multiple growth stages by comparing different 3D data collection methods.…”

Nondestructive Determination of Nitrogen, Phosphorus and Potassium Contents in Greenhouse Tomato Plants Based on Multispectral Three-Dimensional Imaging

“…Due to the complex three-dimensional (3D) morphology of plants, especially for the large plants, it is difficult to ensure measurement accuracy and stability when the characteristics from single-view 2D plant canopy imagery or single point are used to establish models for determining plant nutrients. Methods for the 3D reconstruction of plants are relatively mature and are mainly based on 2D laser lidar [39], 3D laser lidar [40], Kinect sensor [41], multiview stereo reconstruction [42], and multicamera synchronous reconstruction [43] technologies. However, the reconstructed plant 3D point cloud model contains only coordinate information and RGB color information.…”

“…A Kinect sensor pose estimation and self-calibration method were established, enabling the unified transformation of the coordinate system of the multiview point cloud and the rough registration of multiview point clouds. Finally, the iterative closest point (ICP) algorithm was used for the precise registration of multiview point clouds [41,47], thus reconstructing a multispectral 3D point cloud model of tomato plants. The reflectance information of the multispectral 3D point cloud model of the tomato plant canopy was the input value, and chemical measures of the plant NPK contents were the output values.…”