Abstract: Based on a quadratic measurement error model，an efficient computational method for estimating the fundamental matrix is presented.At first，to improve the numerical stability，the matching point pairs in each image are normalized via translation and scale regulation，then the Golden Section method is used to complete minimization of a non-smooth function and to determine the level of noise, and the fundamental matrix is then achieved.Numerical experimental results show the effectiveness of the proposed method.

Key words: fundamental matrix, quadratic measurement error model, normalization, golden section