Holography is a technique to record and reconstruct a three dimensional scene on a two dimensional plane based on optical diffraction and interference. The full three- dimensional object information can be recorded in a fringe pattern image known as a hologram. In modern times, a digital hologram can be acquired, processed, stored and reconstructed all in digital forms. Various hologram image processing techniques can be developed with regard to the specific image properties of digital holograms. In this thesis, the author explores two novel hologram image processing issues, i.e., hologram decomposition and hologram inpainting.
By hologram decomposition, one hologram can be decomposed into several sub-holograms and each sub-hologram represents one individual item in the 3D object scene. In this thesis, a Virtual Diffraction Plane (VDP) based hologram decomposition scheme is proposed. By Otsu thresholding segmentation, morphological dilation and sequential scan labelling, a VDP back propagated from hologram can be divided into sub-VDPs and each sub-VDP can be reverted to be a sub-hologram. At the same time, an alternative VDP cutting approach is proposed as a supplementary method. This automatic hologram decomposition scheme is further employed for focus distance detection in blind hologram reconstruction. It is free of “Shading Effect” and has better accuracy and reliability.
By hologram inpainting, a damaged hologram can be restored by filling in the missing pixels. In this thesis, an exemplar and search based image inpainting technique is applied for the restoration of heavily damaged digital off-axis Fresnel holograms. An intelligent search technique, Artificial Bee Colony algorithm, is employed to significantly increase the computation efficiency of the inpainting process.
Experimental results reveal that with this proposed method, the pictorial information in heavily damaged holograms can be recovered with good fidelity. In addition, Morphological Component Analysis (MCA) inpainting algorithm is proposed for hologram restoration when the missing regions are noise-like separate missing pixels instead of scratch lines or holes. Three potential application cases of hologram inpainting are discussed including occlusion recovery, hologram watermarking and QR code restoration.