Canopy Density Estimation of Apple Trees. Issue 32 (2022)

• Record Type: Journal Article
• Authors: Paudel, Achyut; Karkee, Manoj; Davidson, Joseph R.; Grimm, Cindy
• Is Part Of: IFAC-PapersOnLine. Volume 55:Issue 32(2022)
• Page Start: 124
• Page End: 128
• Publication Date: 2022
• Subjects: Canopy density -- canopy segmentation -- agricultural automation -- machine vision -- canopy characterization; Automatic control -- Periodicals; 629.805
• DOI: 10.1016/j.ifacol.2022.11.126
• Languages: English
• ISSNs: 2405-8963

Abstract

Abstract: Tree canopy density is an important parameter in developing a decision support system for precision orchard management including application of the right amount of nutrients at the right time and right location. Previous studies mostly focus on canopy characterization using the light detection and ranging (LiDAR) sensor which lacks critical color and texture information. This study utilized a ground-based stereo-vision sensor mounted on a utility vehicle to capture the canopy data during the growing stage (July, 2021) in a commercial orchard. The acquired color images, along with the point-cloud data, was used to segment out individual trees and estimate canopy density which is the measure of canopy cover per unit total area for each tree. A K-means segmentation method followed by depth thresholding was used to segment the desired tree canopies. The segmentation was compared to the manual segmentation and a F 1 score of 0.78 was obtained. The density was obtained using the ratio of pixel count of vegetation and total area of interest around the trunk of specific trees. The obtained result was compared against the expert's assessment of tree vigor (categorical variable with values from 1 to 5), which showed a good correlation ( R 2 = 0 . 81). The obtained canopy density, along with other parameters including trunk size, and canopy color change during fall, will be used in the future to develop a decision support system for the assessment of nutrient requirement for individual trees to achieve the plant level precision nutrient management.