Most of the cities all over the world have the problem of traffic jam, but Beijing is probably one of the typical cities. Researchers and government have applied the energies to finding a solution to solve this problem in Beijing, but the effect is not satisfying. Space syntax theory and its mode to solve problems can provide instructive reference for analyzing the urban traffic problem. Based on the current condition of urban traffic network in Beijing, this paper was in virtue of space syntax theory and GIS visualization technology, extended the space syntax theory, divided the traffic network according to roads' designation, and improved the axial map. We put forward the concept of accessibility, discussed the method that how to deal with loops, and proved the optimizing measures which are supposed to revive the current traffic pressure in Beijing based on the iterative optimizing experiments. It is concluded that increasing radialized roads can help to improve the traffic status obviously of Beijing traffic network, which is in the framework of loops. Moreover, we conducted different simulating experiments, such as adding 2 more, 4 more, 6 more, 8 more, and up to 14 more radialized roads in different directions. We discovered that there is a critical of the added roads. Adding 6~8 more roads can lead to an optimal yield. In accordance with this finding, we selected the axial lines of roads on the spot according to current urban planning, and had validated the feasibility and rationality of the optimizing method. It is concluded that building some new radialized roads and rebuilding some main roads are available for improve the traffic condition and the input-output ratio is better. Consequently, the extended space syntax method can instruct road reconstructing and urban planning, and be generalized to other cities.