Page 411 - NGTU_paper_withoutVideo

Modern Geomatics Technologies and Applications

Fig 6: bias histogram for crop row or a soil row.
However, after classifying the lines based on the bias, these discrete lines must be merged and integrated to obtain the
final crop lines. For this purpose, the lines sorting by x coordinates according to the starting point. Each crop line will have a
first point and an endpoint. The first line's starting point is considered the first point, and the endpoint is considered the potentially
endpoint. Now the distance between the potentially endpoint and the starting and endpoint of the following line is calculated,
and by knowing the approximate value of the distance between the two crop lines in one direction (dh), four case are possible
according to Figure (7).
 First case: if d1 > 0 and d2 > 0, then the third line's endpoint is considered a potential endpoint, and the algorithm
resumes.
 Second case: if d1 < 0 and d2 > 0, then the next line's endpoint is considered a potential endpoint, and the
algorithm resumes.
 Third case: if d1 < Δd and d2 > 0, then the endpoint of the next line is considered as a potential endpoint, and
the algorithm resumes.
 Fourth case: if d1 > Δd and d2 > 0, in this case, the potential endpoint is accepted, and the connection line of
the first point and endpoint is considered as the final line of crop. The starting point of the next line will be the
new first point, and the algorithm will resume again to extract all the lines.

Fig 7: four rules for merging pieces of initial lines

6