Based on plot experiments of winter wheat in Guanzhong area of Shaanxi Province, canopy hyper-spectral reflectance data was collected at different winter wheat growth stages. Seven absorption features were calculated based on continuum removed spectra of original canopy spectrum, which were used to explore the quantitative relationship with leaf nitrogen content. The results showed that continuum removed processing enlarged the small differences inherent in original canopy spectra and absorption characteristics controlled by nitrogen stress. The continuum removal spectra obviously improved the correlationship between original canopy reflectance and leaf nitrogen content. It was negatively correlated with leaf nitrogen content in 721~727nm with determination coefficient of -0.851. The absorption characteristics parameters improved the leaf nitrogen estimation accuracy compared with sensitive bands, and the estimation accuracy by absorption parameters in 550~770nm performed well than that by the variables in 400~550 significantly. Between all the absorption parameters in 550~770nm and 400~770nm bands, the total absorption area, left absorption area, right absorption area and area normalized maximal absorption deep gave more stable achievements than the others, and passed the significant test at P<0.01 level with absolute value of correlation coefficient over 0.8. The best prediction model was built by total absorption area in 550~770nm with determination coefficient of 0.82 with exponential formation. Accuracy test also confirmed the conclusion that the prediction model by total absorption area in 550~770nm was more suitable for estimating the leaf nitrogen content of winter wheat quantificationally.