Evaluation of Nutritional Status of Calcium in Corn Plants Using Artificial Vision System.

Nutritional deficiency in plants is identified by chemical analysis of leaf tissue and/or visual diagnosis. Chemical analysis occurs in advanced development stage while the visual diagnosis is difficult to be detected. The artificial vision system (AVS) is a technology that allows identifying nutritional deficiencies in plants independently of human evaluation. This study aimed to produce plants with levels of calcium (Ca) and evaluate the processing of digital images by AVS to identify deficiencies of Ca in corn (Zea mays L.). The experiment was carried out with corn in hydroponic condition with nutrient solution of Hoagland and Arnon (1950) in greenhouse in USP/FZEA/Brazil. The treatments were 0, 33, 66 and 100% of Ca in solution, in a completely randomized design, 4 replicates. The 9600 DPI scanner was used to scanning images of leaf 4 (F4) in V4, leaf 6 (F6) in V6 and new leaf (NL) in V4 and V6. The images were processed by AVS and the leaves were analyzed chemically. The AVS technique evaluated was Gabor wavelets multispectral and fractals using color images (GWC). The base, middle and tip of the leaves were analyzed. The increase of Ca in the nutrient solution led to increased to concentration of Ca in the NL scanned, according to models: y=-0.00002x²–0.0002x+0.65 (R²=0.95) in V4 and y=-0.00002x²+0.008x+0.51 (R²=0.99) in V6. The AVS identified the deficiency of Ca showing 81.1% and 71.6% of accuracy for the NL tip, respectively in V4 and V6; and 73.4% of accuracy in F4 e 55.9% in F6. The best results were in V4, which is very good for the correction of nutritional plants. The GWC of AVS was efficient in  processing digital images and it was better identified the isolated deficiency of Ca in NL of corn leaves in V4.