The invention discloses a stereo image quality objective evaluation method based on multi-scale binocular fusion and local feature extraction. The method comprises the following steps: 1, synthesizing a to-be-tested stereo image pair and a corresponding reference image pair into one image through a disparity estimation and binocular fusion method; 2, carrying out the feature extraction of the synthesized image in a local binary mode; 3, taking the similarity of the features of two sides as the input, taking the quality score as the output, employing a support vector machine for training, and obtaining a mathematic model which can achieve the objective evaluation of the quality of the stereo image according to the known subjective quality evaluation score of the image. The method achieves the efficient and accurate extraction of the features, related with the image quality, of the stereo image, and achieves a function of objectively evaluating the quality of the stereo image. Moreover, a stereo image quality evaluation result obtained through the method is very consistent with a subjective evaluation result of a person through eyes.