To eliminate the artifacts in X-ray projection images and improve the quality o f the three-dimensional reconstruction images using X-ray projection images,a modif ied BSCB inpainting algorithm is proposed based on the characteristics of scintillation defects in the projected images.The Kirsch operator is operated on a normal image for extra cting the mask image,on which the defective pixels of the image are marked.On this basis,corresponding defective pixels in the projec tion images are inpainted and diffused from outer to inner layer.After the inpainting of the outer region is accomplished,the inner region will be inpainted by using the information of the outer region which has been inpainted.And the iteration times and coefficient are determined by the defective pixels around.Th e inpainting process is accomplished by using log operator and the filtered information is diffused alon g the isolux line.BSCB and IBSCB algorithms are evaluated from both subjective vision and objective dat a.Numerical results show that under the different magnifications,the IBSCB algorithm provides an imp rovement in inpainting quality.Also,the IBSCB algorithm is five to six times faster than the BSCB algorithm. Three-dimensional reconstruction images using X-ray projection images before a nd after restoration are compared,which further verifies the effectiveness of the proposed algorithm.