Nondestructive detection for wood is more difficult than that for other materials because of characteristics of wood itself. According to the characteristics like complex internal structure and wide density differences among different tree species, a fast algorithm of inverse projection coordinates was proposed. A wood nondestructive detection and CT imaging system can be structured on the basis of this algorithm. This system consists of an Xray emitter, an offset plate detector, a rotating carrier platform, a data acquisition and imaging software. Setting glued laminated timber and small logs as test objects, the feasibility of the algorithm was verified. Firstly, the projection data were collected and corrected, then the projection coordinates can be calculated according to the distance from the source to the center of rotation, and the filtering function was set up and used to calculate the convolution operation with projection data. The wood faultage image was reestablished at last. The result shows that this algorithm can shorten the image reconstruction time of CT system to less than 1s. Its image resolution can reach 0.052mm2, and it can also show the internal defects such as holes and cracks, and structural features like wood knots, rings, changes between early wood and late wood, which verify the validity of the algorithm applied to wood nondestructive testing and CT imaging system. This system provides a effective method for wood nondestructive detection.