Abstract: This paper mainly aims at the simulation study of high-density-difference multilayer spherical object reconstruction from angle-sparse projection data based on FBP algorithm. Establishes detection object model reference to the material and structure feature of the special workpieces, and develops parallel beam CT reconstruction simulation software platform based on FBP algorithm with C++ language. Uses the direct-zero and close-to-fill two methods to fill the missing sparse projection data, and carries on the simulation reconstruction and comparative analysis. Researches on the image enhancement processing method in a targeted manner, the defect feature highlights. The research contents and methods of thesis can provide technical reference to nondestructive CT tests which have similar characteristics.