Research On Algorithms Of Computer-generated Hologram For Three-dimensional Object Based On Tomography

The unique advantages of computer-generated hologram(CGH)in optical and related fields make it possible for real three-dimension(3D)display.However,there are still some difficult problems in the 3D algorithms of CGH because real 3D scene has large amount of data,wide field of view and complex color information.In this thesis,we study the CGH algorithms for 3D object,and fast generate hologram of true color 3D object and obtain high quality reconstruction images.First,the basic principle of CGH is studied in this thesis.3D image data of real people are obtained by using 3D scanner.The hologram of 3D objects is generated by using point source method in MATLAB,and the numerical reconstruction image is carried out.Then,we use MATLAB to call Graphics Processing Unit(GPU)to calculate the hologram of the 3D image.It is found that the calculation speed of the hologram is about 6.6 times higher after calling GPU.Secondly,we study the basic principle of computer generated Fresnel hologram of a 3D object based on intensity map and depth map.And the S-FFT algorithm,T-FFT algorithm and D-FFT algorithm of Fresnel diffraction are studied in the process of tomography.Through the comparison of reconstruction images and computation time from these algorithms,we find that D-FFT algorithm has advantages in computing speed of hologram,quality of reconstruction images and complexity of algorithm.Finally,we make further studies on angular-spectrum layer-oriented method and the Gerchberg–Saxton(GS)algorithm.The holograms and reconstruction images of true color 3D objects are calculated by using angular-spectrum layer-oriented method.To solve speckle noise of the numerical reconstruction holograms,we introduce the GS algorithm and propose the GS algorithm based on angular-spectrum layer-oriented method.Under the same conditions,we use the proposed algorithm and the traditional angular-spectrum layer-oriented method to calculate the same true color object,respectively.And the root mean square error(RMSE)and peak signal-to-noise ratio(PSNR)of reconstruction images are calculated.It is found that the algorithm proposed in this thesis can successfully suppress speckle noise and improve the quality of the numerical reconstruction images considerably.