The invention discloses a quality evaluation method for a relocated stereo image. The method comprises the following steps: calculating the relative width characteristics of all occluded pixels in a disparity image between left and right viewpoint images of a relocated stereo image, as well as the depth difference characteristics, disparity magnitude characteristics and disparity gradient characteristics of all matched pixels, and calculating the length-width ratio similarity of the left viewpoint image and the length-width ratio similarity of the right viewpoint image to obtain a characteristic vector of the relocated stereo image; training characteristic vectors in a training set by using support vector regression in a training stage; and obtaining the objective quality evaluation prediction value of the relocated stereo image corresponding to the characteristic vector in a test set by using a constructed support vector regression training model in the testing stage. The obtained characteristic vector has relatively high stability and can well reflect the depth, comfort and image quality change of the relocated stereo image, thus effectively improving the correlation between an objective evaluation result and subjective perception.