Rapid detection the content of chlorophyll a, chlorophyll b, β-carotene and lutein on fresh cucumber leaves by using hyper spectral image and high performance liquid chromatography (HPLC) was done. One hundred and twenty hyper-spectral images of cucumber leaves were collected by near infrared hyper-spectral camera, then using HPLC to accurate detect the content of chlorophyll a, chlorophyll b, β-carotene and lutein. After hyper-spectral images have been corrected, a 50×50 pixels region as region of interest (ROI) was defined. Average spectrum from ROI was extracted and prediction model of partial least squares (PLS) with the content of four pigments was built. In order to advance the stability and prediction accuracy, interval partial least squares(iPLS) , backwards interval partial least squares(BiPLS) and synergy interval partial least squares(SiPLS) algorithm was used to select diagnostic bands of each pigments after optimized the number of spectrum interval. The result showed that using BiPLS and SiPLS would get the best PLS model, the best prediction coefficient of chlorophyll a, chlorophyll b, β-carotene and lutein is 0.8257, 0.8134, 0.8116 and 0.8262, respectively.