A “sample independence” restoration method aiming at deep layer attenuation correction of three dimensional confocal scanning images is given with the analysis of its mathematical models presented in details. A rapid access based on FFT is derived with the presumosition that the resulted images will be precise enough. The algorithm, without additional noise inducement, is also proved to be accurate, efficient and flexible. The computing efficiency and memory consumption are discussed followed by its computational realization. Approaches for better performance are proposed subsequently. The successful applications to both test images and real ones are also given.