The hyperspectral images,SPAD and chlorophyll content from different part of leaf of different rapeseed plants under different nitrogen levels were collected.The methods of preprocessing hyperspectral data were calculated and compared.The PLS (partial least squares) and LS-SVM (least squares support vector machine) methods were used to build prediction model of rapeseed leaf SPAD.The results showed that the result of LS-SVM prediction model based on the SNV (standard normal variate) processing method named as SNV-LS-SVM was the best.The map of rapeseed leaf spatial distribution SPAD was constructed according the best prediction model.The SPAD value extracted from SPAD map of different rapeseed plants and positions was analyzed with chlorophyll content.The results showed that leaf on the top part of fourth rapeseed plant was the optimal measurement position.This paper combines the advantage of hyperspectral imaging technique and the non-destructive SPAD measurement method.The methods proposed realized the predicting SPAD spatial distribution of rapeseed leaf and the optimizing measurement position identification.It will provide a theoretical basis and methodological guidance for efficiently detecting chlorophyll content in rapeseed plant.