The invention discloses a no-reference stereo image quality evaluation method based on a quality map generation network. The method comprises the steps of in a training stage, constructing a quality map generation network, sequentially inputting normalized images of all distorted plane images in a training set into the quality map generation network, taking a corresponding quality map as supervision, and performing training to obtain an optimal quality map generation network model; Test phase, utilizing the optimal quality map to generate a network model to extract a left viewpoint image of the distorted stereo image to be evaluated; predicting quality maps of the normalized images of the right viewpoint image and the fused image respectively, acquiring respective salient feature maps, solving respective predicted quality scores according to the corresponding predicted quality map and the corresponding salient feature map, and fusing the three predicted quality scores to obtain a finalpredicted quality score; The method has the advantages that the influence of various characteristics of the stereoscopic image on the visual quality can be fully considered, so that the correlation between an objective evaluation result and binocular subjective perception can be improved.