The invention discloses a traffic monitoring image rain removing method based on anisotropic sparse gradient. A mathematical model is shown in the specification, wherein f is a degraded image shot inrainy days, u is a clear image to be restored, v is a rain layer image to be detected, [delta]<x> and [delta]<y> are gradient operators in the vertical direction and the horizontal direction of the image respectively, ||.||<L0> is an L0 norm for counting the number of non-zero elements, and the ||[delta]<x>u||<L0> and the||[delta]<y>u||<L0> are vertical and horizontal gradient regularization termsof the image, and ||v||<L0> is a rain detection regularization term, wherein alpha<1>, alpha<2>, beta<1> and beta<2> are non-zero regularization coefficients, the alpha<1>, the alpha<2> and the beta<1> are image horizontal and vertical and rain information regularization coefficients respectively, and the beta<2> is a regularization parameter of a fidelity term. The parameters alpha<1>, alpha<2>,beta<1> and beta<2> can be adjusted through experiments, and a more effective rain removal effect is obtained. According to the invention, the restored traffic monitoring image not only has better edge structure information, but also can effectively measure the position of rain in the image.