For the effects of atmospheric turbulence on Acquisition, Tracking and Pointing (ATP) of the beacon in laser communication, a rapid restoration and real-time detection algorithm based on blind deconvolution was proposed. A local degradation model was proposed based on one dimensional Point Spread Function (PSF) instead of the normal two dimensional model.Then the constraint operator of the classic constrained conjugate gradient algorithm was improved, and the original image was estimated by the improved conjugate gradient iterative algorithm. Finally, the centroid of the beacon was calculated from the estimation image by connected component analysis. To meet the real-time requirement, a Field Programming Gate Array( FPGA) and Digital Signal Processors(DSPs) were used to realize the proposed algorithm and to extract the central position of a spot . Experimental results indicate that the real-time restoration on the image with 200×200 pixels and 100 Hz frame rate have been obtained by the proposed algorithm. The error between the real-time result and the after one calculation is less than 1 pixel, which meets the requirements of the laser communication system for real-time tracking on the beacon.