To study the effect of spectral reconstruction algorithm on spectral reflectance reconstruction accuracy and chromaticity in order to meet the demand of image reproduction to a greater extent and avoid the phenomenon of isochromia. An inverse variance weighted regression spectral reflectance reconstruction algorithm was proposed by improving the traditional principal component analysis method. Firstly, the inverse variance weighting model is introduced to adjust the weight of factors. Secondly, the spectral reflectance training sample set is normalized and a new covariance matrix is calculated. Finally, the spectral reflectance is reconstructed by singular value decomposition of the weighted spectral reflectance sample set. Experiments show that the color accuracy of CIE DE2000 based on inverse variance weighted spectral reflectance reconstruction algorithm is increased by 3.3%, spectral reconstruction accuracy is increased by 2%, and good color reproduction effect can be achieved. The weighted algorithm can be used in spectral reflectance reconstruction, which is helpful to improve the accuracy and chrominance of spectral reconstruction.