To solve the problems of inherent errors caused by the residual stray light after the stray light limitation of the optical systems as well as time-varying errors caused by system aging and improve the accuracy of the optical measurement system, an error correction algorithm is proposed for a photometric measurement system based on the radial basis function (RBF) network. In this study, the point source transmittivity is used to analyze the measurement error distribution of the photometric measurement system. Further, the error correction algorithm is designed and improved based on the RBF network in accordance with the approximate estimation of stray light distribution. TracePro software is employed for comparative simulations. The results demonstrate that the photometric measurement error can be reduced to less than 0.24% by correcting the error compensation network. Compared with the general RBF network, the convergence speed and approximation ability of the proposed algorithm are obviously improved, providing a more rapid and effective tool for solving the problems entailed by the stray light limitation.