Title : Real-time lidar data acquisition and 3d image reconstruction for orchard canopy profile measurement
Abstract:
The ability to quickly measure and represent tree canopy profile in three dimensions (3D) is extremely important for precision agriculture applications. In this paper, we proposed a laser scanning sensor-based object reconstruction algorithm to rebuild and display 3D plant target images in real-time and compute geometrical characteristics of tree canopy profile. The developed algorithm can process 2D laser sensor raw data to build a pseudo-color 3D image in real time and to measure the dimensions of target tree canopy profile. The developed software program was written in C++ based on OpenCV library. It also integrates user-friendly GUI to provide user machine interactions. The software program was tested on a specially designed indoor LiDAR-based canopy profile measurement platform. Two artificial trees (tree-1 and tree-2) were used as target objects. The reconstructed 3D image and data analysis results showed that the performance of the developed algorithm to measure tree canopy profile and to reconstruct the 3D image in real time is very well at travel speed ranged between 1.0m/s and 2.0m/s and detection distances ranged from 1.5m to 2.0m. The mean RMSE values for height measurement in tree-1 and tree-2 were 72.5mm and 17.8mm respectively. In width measurement, the mean RMSE values were 50.7mm for tree-1 and 80.9mm for tree-2. For a real filed application, this study plays an initial level role in several aspects including generating real time control signals based on target plantgeometrical characteristics for controlling and monitoring processes in orchard and crop management. Key words: precision agriculture, LiDAR range finder, 3D plant image, real-time reconstruction, canopy measurement OpenCV, and Qt creator.
