An improved reconstruction algorithm of the boundary surfaces within 3D images was proposed based on the fractional differential, which can reconstruct more detailed structures of the 3D irregular biomedical objectives. Thus, the deficiency of reconstruction more detailed information in the was proposed based on the fractional differential, which can reconstruct more detailed structures of the 3D irregular biomedical objectives. Thus, the deficiency of reconstruction more detailed information in the traditional algorithm was overcome. According to the separateness of the Fourier Transformation of the fractional differential, the 2D calculus was extended to 3D, and the 3D discrete templates were deduced. Due to the nonlinear enhancement, the 3D fractional differential can enhancement the 3D slice data significantly. By comparison with the traditional reconstruction method of the 3D boundary surfaces, the improved algorithm could detect and extract more detailed 3D structure information by using the 3D fractional differential. The proposed algorithm was applied in the reconstruction of the 3D irregular structure of the neuron within confocal microscopy images, which demonstrated the high efficiency and property, it can be easily extended to other research fields of 3D visualization.