The invention discloses a singular value decomposition(SVD)-based image quality evaluation method, which comprises the following five steps of: 1, respectively dividing an original image and a distorted image with the size of m*n into m*n/k*k image blocks; 2, programming all the image blocks according to an SVD principle so as to implement SVD; 3, calculating structural distortion in left and right singular vectors U and V and U(p) and V(p) of the image blocks of the original image and the distorted image, and simultaneously calculating brightness distortion in singular value characteristic vectors S and S(p) of the image blocks of the original image and the distorted image; 4, integrating the brightness distortion and structural distortion of the image blocks of the original image and the distorted image to obtain a formula D1, performing cycle calculation to obtain D1 values of all m*n/k*k image blocks, and solving a mean value; and 5, integrating the image quality evaluation parameters, defining an SVD-based image quality evaluation index, wherein when a QSVD value is zero, the image quality is the best, the QSVD value is increased with the change of QSVD, and the image quality is worse. The SVD-based image quality evaluation method has the advantages of practical value and wide application prospect in the field of image quality evaluation.