Detection of crop rows and drip irrigation lines using UAV imagery in agricultural fields

Agricultural irrigation constitutes a significant portion of water consumption worldwide. Drip irrigation systems enable the efficient use of water in agricultural irrigation and contribute to water conservation. The correct placement of drip irrigation lines is critical to ensure that plants receive an adequate amount of water. Checking the correct placement of drip irrigation lines in large-scale fields is quite time-consuming. In this thesis, a novel method is proposed to detect plant rows and drip pipelines on images obtained with an unmanned aerial vehicle. A new dataset was created with images taken from cornfields using an unmanned aerial vehicle. On the images in this dataset, color thresholding and various morphological image processing methods were applied to find crop rows, which is the first step of the proposed method. Then, the line segments obtained were combined based on their angles and distances to identify the entire line of the crop row. After detecting the crop rows, various image preprocessing methods and the relationships between neighboring line segments were utilized to identify drip irrigation lines in the areas between the crop rows. Experimental studies separately evaluated the performances of the algorithms for crop row detection and drip irrigation line detection. The proposed method demonstrated performance with 99.01% and 86.54% accuracy for the two problems mentioned above, respectively.