Full text
| |
| Author(s): |
Fernanda de Fátima da Silva
Total Authors: 1
|
| Document type: | Doctoral Thesis |
| Press: | Pirassununga. |
| Institution: | Universidade de São Paulo (USP). Faculdade de Zootecnica e Engenharia de Alimentos (FZE/BT) |
| Defense date: | 2015-04-30 |
| Examining board members: |
Pedro Henrique de Cerqueira Luz;
Telmo Jorge Carneiro Amado;
Murilo Mesquita Baesso;
Valdo Rodrigues Herling;
Marcio Roberto Soares
|
| Advisor: | Pedro Henrique de Cerqueira Luz |
| Abstract | |
The fertilization and the use of more productive cultivars are important technologies to improve productivity and sustainability of the maize crop. The digital image analysis is a technology used to identify nutritional deficiency in corn leaves, on the methodologies currently being applied, the correction of the deficient nutrient is not possible in the same growing cycle. The objective of this study was to evaluate the nutritional and productive characteristics and verify the extraction methods of digital image characteristics. This is performed to diagnose symptoms of nitrogen deficiency (N) and potassium (K) in corn hybrids (Zea mays L.) grown in a greenhouse, with induced nutritional deficiency in nitrogen (N) or potassium (K); and thereafter grown in the field. The experiment was independent for each element and conducted in two steps: 1; in a greenhouse under hydroponic cultivation, with treatments in a factorial 4 (doses) x 3 (hybrid) and 4 replications, with four doses: 5, 20%, 100% and 200% of full dose; and 2; the field in randomized blocks in a factorial 4x3 (4 doses and 3 hybrids) and 4 blocks, 4 fertilizer levels: individual and complete omission (0%) of N or K, 50%, 100% and 200 % of the recommended dose of the nutrient under study. The hybrids were DKB390 PróR2® (H1), Pioneer 30F35® (H2) and Syngenta Status® (H3). The collection and digitalization of the leaves were realizated in V4 and R1 growth stages. Indicative leaves of each stage were obtained and processed by image analysis and chemically analyzed. The extraction methods of characteristics based on texture patterns of images in gray-scale were: Fourier Descriptor Fractal (Fractal), Local Binary Pattern (LBP), Gabor Wavelets (GW) and Gabor Wavelets + Fractal Descriptors (GWF); and also characteristics extraction methods were studied based on 4 spectral indices of color images: excessive green (EVD), normalized red (Vern), normalized green (Vn), green-red ratio (RVV) and the combination of them. In the greenhouse were determined the dry mass of the aerial part of the plant (MSPA) and root (MSRz) and the concentrations of macro and micronutrients. In the field, at the end of the cycle, evaluations were carried out in productivity and foliar and soil analysis. The reduction in N or K levels in the studied hybrids led to significant decreases in foliar nutrient concentration in the plants conducted in a greenhouse and in the field in two stages evaluated, with typical visual symptoms of N deficiency or K for hybrid conducted with the lower dose of this nutrient. The MSPA and MSRZ in hybrid conducted in a greenhouse and productivity in the field were also committed to reducing the availability of N or K for the studied hybrids. To the hybrids conducted in the greenhouse with doses of N or K, when studied separately in the texture based methods, the Fourier showed high percentage of correct answers in all hybrids in both stages, except for H3 in R1. The best methods based on spectral indices presented classification considered more than good. (AU) | |
| FAPESP's process: | 11/20631-0 - NUTRITIONAL STATUS OF MAIZE HYBRIDS BY ARTIFICIAL VISION SYSTEM SUBJECT TO LEVELS OF NITROGEN AND POTASSIUM. |
| Grantee: | Fernanda de Fátima da Silva Devechio |
| Support Opportunities: | Scholarships in Brazil - Doctorate |