Evaluating the suitability of hyper- and multispectral imaging to detect foliar symptoms of the grapevine trunk disease Esca in vineyards

Bendel, Nele; Kicherer, Anna; Backhaus, Andreas; Klück, Hans-Christian; Seiffert, Udo; Fischer, Michael; Voegele, Ralf T.; Töpfer, Reinhard

doi:10.1186/s13007-020-00685-3

Plant Methods

Table 4 Results of the multispectral simulation for the three differentiation tasks

From: Evaluating the suitability of hyper- and multispectral imaging to detect foliar symptoms of the grapevine trunk disease Esca in vineyards

			VNIR												SWIR
			HSI	Multispectral										HSI	Multispectral
			186	2	3	4	5	6	7	8	9	10	RE	288	2	3	4	5	6	7	8	9	10
2016	Symptomatic (original)	CA (%)	73	60	66	71	72	74	75	75	76	76	70	73	58	60	63	65	66	67	68	69	70
		TPR (%)	71	77	71	70	72	72	79	77	77	78	71	70	66	66	66	65	69	66	73	71	74
		FPR (%)	29	37	39	27	26	25	26	26	24	24	30	34	50	40	35	35	34	35	33	30	32
	Symptomatic (annotated)	CA (%)	92	68	69	79	84	86	91	90	91	91	82	88	61	85	85	86	86	86	87	87	87
		TPR (%)	89	69	72	78	83	86	88	88	88	88	86	86	84	71	71	75	76	77	79	81	81
		FPR (%)	0	32	31	19	13	10	6	6	7	6	22	2	48	1	1	3	3	6	4	7	6
	Pre-symptomatic	CA (%)	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–
		TPR (%)	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–
		FPR (%)	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–	–
2017	Symptomatic (original)	CA (%)	70	57	61	66	70	70	71	72	72	73	68	81	62	66	71	74	78	78	79	81	83
		TPR (%)	73	80	68	74	74	72	73	78	76	77	72	82	81	77	75	80	84	82	81	84	86
		FPR (%)	32	45	42	39	33	32	32	31	30	30	35	26	44	43	34	31	27	27	23	22	18
	Symptomatic (annotated)	CA (%)	90	67	77	73	85	85	86	87	86	89	83	95	65	66	71	77	79	84	87	88	89
		TPR (%)	90	74	76	81	87	89	89	89	88	91	84	90	77	77	79	78	78	86	86	89	87
		FPR (%)	11	41	22	34	16	17	15	14	15	13	19	6	44	33	31	24	22	16	12	11	11
	Pre-symptomatic	CA (%)	62	53	54	56	59	60	61	62	63	65	59	74	57	62	65	66	67	68	70	71	71
		TPR (%)	63	38	54	48	61	65	66	60	56	64	56	78	68	71	71	70	77	73	81	77	77
		FPR (%)	35	36	39	36	44	36	32	33	34	31	37	32	52	45	40	41	39	36	36	35	36
2018	Symptomatic (original)	CA (%)	77	63	63	73	75	75	76	76	77	77	68	80	60	60	68	70	72	75	75	77	79
		TPR (%)	72	74	66	73	76	74	74	77	77	80	67	74	56	60	64	67	68	74	78	78	82
		FPR (%)	22	41	33	35	25	22	23	25	24	23	31	20	41	37	27	30	28	23	25	21	20
	Symptomatic (annotated)	CA (%)	94	71	80	81	82	83	89	90	91	91	84	92	67	73	76	79	85	87	87	88	89
		TPR (%)	93	71	81	82	82	85	89	89	91	90	83	100	68	65	73	72	83	84	86	87	88
		FPR (%)	5	25	21	20	17	16	11	10	10	10	16	5	28	17	17	16	12	10	12	10	9
	Pre-symptomatic	CA (%)	79	62	66	68	70	71	74	74	75	76	69	86	60	63	65	75	77	78	80	81	82
		TPR (%)	83	77	73	78	82	82	78	80	81	80	72	87	54	71	63	78	80	80	82	84	79
		FPR (%)	28	41	35	39	24	35	31	31	29	29	34	21	42	40	42	24	22	22	19	17	19

CA classification accuracy, TPR true-positive rate, FPR false-positive rate, HSI Hyperspectral Imaging, RE simulation RedEdge™ wavelengths, 186/288 = wavelength bands of the hyperspectral sensors, 2–10 = number of selected spectral bands

Back to article page

ISSN: 1746-4811

Contact us

Submission enquiries: journalsubmissions@springernature.com