According to the spectral features of domestic ZY-3 remote sensing images, the formula of LBV transformation for ZY-3 is proposed and deduced, and the feasibility of improving the quality of ZY-3 remote sensing images is testified. At first, based on the characteristics of ZY-3 remote sensing images, the spectral information of nine types of typical ground features were selected, and regression coefficients were used to calculate regression coefficients. Then, the three components of L, B, V of ZY-3 satellite images were calculated according to the characteristics of the typical ground features space (bare land, water body, vegetation), color space (red, green, blue) and the space of LBV variables (the general radiance level of the ground objects, the visiable - infrared radiation balance, the band radiance variation vector). Finally, the experiments of ZY-3 remote sensing image in Ningde City of Fujian Province were carried out, and quantitative analysis was conducted to evaluate the experimental results. Firstly, the results show that, in the aspect of the visual effects, compared with the original image, the transformed image is more clear, and the details are more abundant, and thus can contribute more to the determination and identification of subsequent features. Secondly, through the LBV transformation, the image information entropy is 6.21, the average gradient is 4.71, the deviation coefficient is 0.46, and the quality of the remote sensing image is better than other transformation methods. Thirdly, by classifying the LBV image, the overall accuracy is up to 89.71%, and the Kappa coefficient is the highest, reaching 0.875 3. The classification accuracy is higher than that of other transformation methods. Therefore, The LBV transformation can improve the quality of ZY-3 remote sensing image, and it can be applied to ZY-3 remote sensing image processing and information extraction.