AN IMAGE PROCESSING APPROACH FOR MONITORING SOIL PLOWING BASED ON DRONE RGB IMAGES

Abstract

Soil tillage is a crucial stage in growing plants. Plant roots need soil cavities with good aeration which is obtained from an excellent soil-plowing process. Controlling the quality of plowing process should be done quickly and precisely since it affects the planting schedule and seed handling in the field. Monitoring the plowing area using drone is the best way since it has low-cost operations and is easy to operate. Most drones used today are equipped with a CMOS camera sensor that produces RGB images with good resolution. This study tries to maximize these RGB images to analyze the plowing area and plowing depth using the vegetative indices formulas and GLCM function. Vari formula is the best vegetive indices compared with VIgreen and GLI formula that can be used to distinguish plowed and unplowed areas in this study. The segmentation algorithm which was developed in this study can detect the plowing area. Based on the test, the segmentation algorithm can detect the plowed area, and the results have been compared with manual observation. The correlation coefficient (r) between the result of the segmentation algorithm and manual observation is 0.77. The composition of RGB in each pixel influences the algorithm’s performance to distinguish the plowed and unplowed areas. However, the GLCM function is not strong enough to estimate the plowing depth because the correlation coefficient is very weak. © 2017 Elsevier Inc. All rights reserved.