In order to carry out the nondestructive detection of chlorophyll fluorescence parameters of potato leaves, hyperspectral images of 200 regions of interest (RoI) were collected and the average reflectance was extracted. Then qP values of the corresponding areas were measured by the closed chlorophyll fluorescence imaging system of Fluorcam. Sample set partitioning based on joint X-Y distance (SPXY) algorithm was used to divide the samples into calibration set (135 samples) and validation set (65 samples) according to the ratio of 2∶1. Algorithms of synergy interval partial least squares (si-PLS) and random frog (RF) were used to select feature wavelengths, and the partial least squares regression (PLSR) model was established with the selected wavelengths. The results showed that the calibration accuracy of si-PLS-PLSR model was 0.6285, the validation accuracy was 0.6103, and the calibration accuracy of RF-PLSR model was 0.7093, the validation accuracy was 0.6872. The interpretation of potato leaves by RF method was better than that by si-PLS method. Selected wavelengths mainly distributed in the range of 518.72~640.64nm, 650~800nm and 850~1000nm, which reflected that fluorescence emission signal was an important response characteristic of photochemical absorption in potato crops. And it also showed that the photochemical absorption was closely related to chlorophyll content, leaf structure, water content and other attributes. The mapping of qP values distribution in leaves provided an intuitive method for the analysis of photochemical absorption and photosynthesis dynamics of potato leaves. The study provided a way for the evaluation of photosynthetic activity, which could also be applied in monitoring of complex physiological and biochemical dynamic of potato crops.