High-throughput phenotyping of lateral expansion and regrowth of spaced Lolium perenne plants using on-field image analysis

Table 1 Description of the different segmentation methods [colour-based (C); texture-based (T); edge-based (E)] used during the images analysis of top view images of L. perenne plants

Name	Segmentation method	Colour space—channel	Domain	Method (WiT)	Extra tools used
Name	Segmentation method	Colour space—channel	Domain	Method (WiT)	Before	After
C1	Colour threshold	HSV–HSV	Spatial	Threshold		Dilate, fill holes, erode, selection of the largest object
C2	Colour threshold	RGB-B/G	Spatial	Threshold		Dilate, fill holes, erode, selection of the largest object
T1	Texture	HSV-S	Frequency	Fast Fourier transform and cosine filter		Dilate, fill holes, erode, selection of the largest object
T2	Texture	HSV-S	Spatial	Entropy measurement		Dilate, fill holes, erode, selection of the largest object
E1	Edge	HSV-S	Spatial	Gradient magnitude and direction filter (Prewitt)	Gauss filter	Dilate, fill holes, erode, selection of the largest object
E2	Edge	HSV-S	Spatial	Two dimensional convolution	Gauss filter	Dilate, fill holes, erode, selection of the largest object
E3	Edge	HSV-S	Spatial	Prewitt edge detection	Gauss filter	Dilate, fill holes, erode, selection of the largest object
E4	Edge	HSV-S	Spatial	Refine edges based on Prewitt	Gauss filter	Dilate, fill holes, erode, selection of the largest object

ISSN: 1746-4811