The leaf pigment content is an important indicator to characterize crop cultivation substrates nutrient elements and physiological state. Rapid and accurate acquisition of pigment content and leaf position distribution is the basis for precise water and fertilizer management in facility agriculture. Chlorophyll a (Chla), Chlorophyll b (Chlb), Chlorophyll (Chll) and Carotenoid (Caro) at different leaf positions of the tomato seedling stage were used as research indicators. Ten nitrogen concentrations were prepared for the nutrient solution. We picked 1710 slices (285 samples) for VIS-NIR hyperspectral acquisition according to the leaf position. The data were preprocessed by Savitzky-Golay (S-G), standard normal variate (SNV), and multiple scattering correction (MSC). Firstly, the key bands were “roughly” extracted with the competitive adaptive reweighted sampling (CARS) algorithm. Then, the iteratively retains informative variables (IRIV) algorithm to judge the importance of key bands, and “accurately” extract the optimal set of bands by reverse eliminating strong and weak bands. The partial least squares regression (PLSR) models were established. The results showed that: (1) When the nutrient liquid nitrogen concentration was 302.84 mg·L-1, the leaf pigment content was the largest. Moreover, the inhibitory effect of high concentration is higher than that of low concentration. The pigment content in the leaf position showed the distribution law of upper>middle>low. (2) The CARS-IRIV-PLSR algorithm, “roughly-accurately” key band screening strategy was used to extract 4, 4, 10 and 11 key bands for Chla, Chlb, Chll and Caro, and the Rp was 0.772 2, 0.732 1, 0.847 1 and 0.858 7, respectively. (3) Combined with the optimal model pigment quantitative inversion image visual expression, the distribution rules of Chla, Chlb and Chll are consistent, while the distribution rules of Caro and Chll are opposite. This conclusion is consistent with the plants physiological characteristics and measurement results. The hyperspectral imaging technology can realize the nondestructive detection and visual expression of leaf pigment content. It provides data support and a theoretical basis for plant leaf pigment distribution, nutrient deficiency and fertilization decision-making in facility agriculture.