Measuring deficiencies in the nutrition of crops is important because it allow growers to take proper action on time so as to get better development of plants. Several methods have been proposed in the measurement of nitrogen. One of those methods is image processing using color spaces. In this study, five levels of deficiency of nitrogen were induced for the evaluation in tomato seedlings. Color images were taken with a digital camera. These images were processed in Visual C++ making segmentation of the images, and for its analysis they were converted from red, green, blue (RGB) to improved hue, luminance and saturation (IHLS) color space. Luminance, saturation, hue and hue-saturation components on IHLS were proposed in the nitrogen diagnostic. Results from laboratory on the tomato leaves were taken as reference. To evaluate the data, linear regression and variance analysis with Tukey test (p<0.01) were made. Hue was the value that had better correlated laboratory values obtaining R 2 =0.86. Nitrogen estimation by this method in tomato seedlings is fast and economic and it has an advantage in relation to other proposed methods in RGB color spaces that it is less susceptible to changes of illumination.

No abstract