The c canopy taking in before and after grazing term in the production mode of cfield goose was used as the research object. Hyperspectral techniques were used to analyze the spectral characteristics of c canopy leaves in different periods, and a full-band based Principal Component Regression model, Partial Least Squares Regression model and Support Vector Machine regression model were established to propose a fast, convenient and efficient hyperspectral imaging detection technique. The results showed that the nitrogen value of the grazing area was always lower than that of the control area, during the grazing period, the reflectance of the near-infrared spectrum increased, and the red edge position moved to the left. In terms of model establishment, the optimal model was obtained for different grazing periods. The positive set determining coefficient (Rc2), the root-mean-square error correction (RMSEC), the prediction set decision coefficient (Rp2) and the root-mean-square error prediction (RMSEP) were obtained by using SNV-BICA-PCA-PLS in the pre-grazing period. Their values were 0.9136, 0.1750, 0.8910 and 0.1052, respectively. The values of Rc2, RMSEC, Rp2 and RMSEP were 0.9006, 0.0418, 0.8508 and 0.1233, respectively, when they were obtained by using MSC-BICA-PCA-MSC in the post-grazing period. The research results provided support and help for the future agriculture and animal husbandry integration to optimize production management and establish a nitrogen nutrient balance model.