Imaging Wheat Canopy Through Stereo Vision: Overcoming the Challenges of the Laboratory to Field Transition for Morphological Features Extraction

Abstract

Stereo vision is a 3D imaging method that allows quick measurement of plant architecture. Historically, the method has mainly been developed in controlled conditions. This study identified several challenges to adapt the method to natural field conditions and propose solutions. The plant traits studied were leaf area, mean leaf angle, leaf angle distribution, and canopy height. The experiment took place in a winter wheat, Triticum aestivum L., field dedicated to fertilization trials at Gembloux (Belgium). Images were acquired thanks to two nadir cameras. A machine learning algorithm using RGB and HSV color spaces is proposed to perform soil-plant segmentation robust to light conditions. The matching between images of the two cameras and the leaf area computation was improved if the number of pixels in the image of a scene was binned from 2560 × 2048 to 1280 × 1024 pixels, for a distance of 1 m between the cameras and the canopy. Height descriptors such as median or 95th percentile of plant heights were useful to precisely compare the development of different canopies. Mean spike top height was measured with an accuracy of 97.1 %. The measurement of leaf area was affected by overlaps between leaves so that a calibration curve was necessary. The leaf area estimation presented a root mean square error (RMSE) of 0.37. The impact of wind on the variability of leaf area measurement was inferior to 3% except at the stem elongation stage. Mean leaf angles ranging from 53°to 62°were computed for the whole growing season. For each acquisition date during the vegetative stages, the variability of mean angle measurement was inferior to 1.5% which underpins that the method is precise.

SegVeg: Segmenting RGB images into green and senescent vegetation by combining deep and shallow methods

“…12, right). In such conditions, some authors [51], propose to assign the saturated pixels to the most frequently saturated class. However, in our case, this would degrade the segmentation performances since the saturated pixels may belong to the three classes, with however a larger representation of green vegetation particularly with glossy leaves under either clear sky conditions or using flashes.…”

Registration and Fusion of Close-Range Multimodal Wheat Images in Field Conditions

“…This choice was made because of the uncertainty on stereovision performances on the various images containing ears (green ears, yellow ears, tilted ears, …). Concerning stereovision applied to wheat canopy, detailed explanations can be found in [34]. For this study, rectification of left and right images was performed by Bouguet's algorithm thanks to the calibration values extracted using the chessboard [35].…”

“…). Concerning stereovision applied to wheat canopy, detailed explanations can be found in [34]. For this study, rectification of left and right images was performed by Bouguet's algorithm thanks to the calibration values extracted using the chessboard [35].…”

Wheat Ear Segmentation Based on a Multisensor System and Superpixel Classification

“…The images were recorded using a color depth of 10 or 12 bits per pixel then reduced to 8 bits per pixel, because the stereovision and registration open-source libraries need 8-bit inputs. The autoexposure algorithms of the RGB and multispectral devices were adapted to prevent image saturation as suggested by [20]. The ILS was positioned above the cameras.…”