Page 408 - NGTU_paper_withoutVideo

Modern Geomatics Technologies and Applications

Fig 1: Flowchart of the proposed algorithm

2.1. Identify the initial lines
 Find the best threshold to detect crops
Crop lines can be defined as a combination of several small parallel lines. Prior to any line detection method, pre-
processors are generally required to remove unwanted information, such as shadows, soil, or rock. This step aims to implement
the Otsu integrated threshold for binary image generation and background isolation from plants [27, 28]. Different indices such
4
3
2
1
5
7
6
as NGRDI , NGBDI , MNENTI , RGBVI , BGVI , G% , and ExG , etc. have been developed to identify vegetation, which was
examined in [15] all available indices, with the G% index having the highest accuracy (91.9%). In this article, the G% method
was used to identify vegetation and background.

% = ( )
( + + )
In Equation (1) G% binary image of vegetation and background removed, Red, Green, and Blue are the red, green, and
blue bands of the input image.

 Apply opening to bolding crop lines
The threshold stage's output does not provide a clear picture of the vegetation, and there is a gap between these crop lines.
To fill the hole space, the Opening Morphology Operator was used along with a disk kernel. The Opening operator is a
combination of Erosion and Dilation.
∘ = ( ⊝ ) ⊕ ( )
In Equation (2), Image is the image of the previous Kernel output threshold, the 10-size disk kernel.
Th

 Detected straight lines using Hough Transform
Now, to identify the crop lines, it is necessary to extract the edges of the features using edge-finding algorithms. The
Canny Edge Detector Algorithm was used as one of the best edge detector algorithms. The Canny Edge Detector is known for
producing thin edges up to one pixel for direct edges. [27]. Hough conversion is one of the most widely used methods for

1
Normalized green red difference index
2 Normalized green blue difference index
3 Modified normalized green red difference index
4 Red green blue vegetation index
5 Blue/green pigment index
6 Green percentage index
7 Excess green
3