Abstract:Seismic tomography is an important technique for estimating subsurface velocity structure.Unlike medical CT,seismic tomography often uses less enough data because seismic rays fail to uniformly cover the objective area.If a poor initial mode1 is chosen in such case,conventional tomography methods,such as damped least square method,algebraic reconstruction technique etc.can hardly produce satisfaetort' results.To cope with the problem,we put forward a restrained tomography method,which solves least square problem after an inequality restriction is given.The solution of optimized problem under the restriction is deduced in the article.The idea of solving this problem is also applicable to solving other nonlinear restrained optimization problems.The final numerical result and the relevant evalualion prove this method effective and superior in the case of less enough data.What is more,the method has certain noise resistance.