| Full text | |
| Author(s): |
Arnal Barbedo, Jayme Garcia
[1]
Total Authors: 1
|
| Affiliation: | [1] Embrapa Agr Informat, BR-13083886 Campinas, SP - Brazil
Total Affiliations: 1
|
| Document type: | Journal article |
| Source: | PLANT DISEASE; v. 98, n. 12, p. 1709-1716, DEC 2014. |
| Web of Science Citations: | 23 |
| Abstract | |
A method is presented to detect and quantify leaf symptoms using conventional color digital images. The method was designed to be completely automatic, eliminating the possibility of human error and reducing time taken to measure disease severity. The program is capable of dealing with images containing multiple leaves, further reducing the time taken. Accurate results are possible when the symptoms and leaf veins have similar color and shade characteristics. The algorithm is subject to one constraint: the background must be as close to white or black as possible. Tests showed that the method provided accurate estimates over a wide variety of conditions, being robust to variation in size, shape, and color of leaves; symptoms; and leaf veins. Low rates of false positives and false negatives occurred due to extrinsic factors such as issues with image capture and the use of extreme file compression ratios. (AU) | |
| FAPESP's process: | 13/06884-8 - Automatic disease diagnosis in plants using digital images |
| Grantee: | Jayme Garcia Arnal Barbedo |
| Support Opportunities: | Regular Research Grants |