Abstract: Abstract: The rapid development of UAV equipment provides a new remote sensing data acquisition platform. The acquisition of airborne hyperspectral image data generally includes three main steps: image acquisition, image preprocessing and image splicing. The acquired data is obtained by segment registration after wave first, and the data quality is crucial to the generation effect of subsequent orthophoto images. Most of the current studies are based on the direct interpretation of Orthophoto images, and it is impossible to avoid the distortion of Orthophoto images caused by the anomalies caused by the previous data acquisition or data preprocessing. The process of remote sensing data acquisition and transmission in natural environment is interfered by many factors, which results in some errors between the collected data and the actual situation. During data acquisition of hyperspectral imaging system of UAV, image distortion, such as white noise and stripe noise, will be caused due to route change and solar irradiance change, which seriously interferes with the acquisition of aerial images. How to establish effective evaluation indicators to guide the quality interpretation of aerial images is a matter of concern. In order to solve this problem, this study uses the ground object spectrometer (350-2 500 nm) and airborne imaging spectrometer (502.56-903.2 nm) to obtain the canopy spectrum of cotton crops in the study area .The aerial image size is 42 bands× 768 pixel×768 pixel. Combined with the central wavelength of the imaging spectrometer, the spectral data of the ground object spectrometer with the same half wave width are separated for spectral information comparison. Analyze the spectral characteristic positions and amplitudes of typical vegetation, such as green peak, red edge and red valley, to verify the quality of spectral information and ensure the accuracy of spectral information acquisition of reference images. Referring to previous research contents and actual data acquisition results, the main distortion types are locked, and the collected high-quality reference images are sequentially generated into five types samples of different degrees, including white noise, defocus blur, motion blur, spectral smoothing and stripe noise, through digital image processing technology. Each type includes 150 samples and a total of 750 samples, Based on the statistical results of the actual noise samples, a total of 50 noise sample sets (stripe noise and mixed white noise) and reference images were constructed by using morphology and interpolation processing. According to the characteristics of hyperspectral images, 3 categories of 15 indexes for calculating the spatial information, spectral information and spatial spectral composite quality of images covering the band are established. With the help of multiple types of samples with different degrees of distortion, the effectiveness of the indexes is evaluated by using the correlation analysis method. The correlation analysis of the indexes is carried out in combination with the two categories of samples. The results show that the each image quality calculation index proposed in this paper was significantly correlated with the deterioration of image quality (P<0.01). The correlation of all indicators for real noise samples has decreased to varying degrees. Only four indicators, mean absolute error MAE (0.609, P<0.01), mean square error MSE (0.459, P<0.01), relative root mean square error RRMSE (0.502, P<0.01) and overall information fidelity F (-0.471, P<0.01) meet the correlation analysis. The research results can provide reference for the quality evaluation of low altitude airborne hyperspectral image data and the quality analysis and distortion index selection in the image processing process.