In this thesis we discuss and examine the contributions we have made to the field of digital hologram image processing. In particular, we will deal with the processing of numerical reconstructions of real-world three-dimensional macroscopic objects recorded by in-line digital holography. Our selection of in-line digital holography over off-axis digital holography is based primarily on resolution. There is evidence that an off-axis architecture requires approximately four times the resolution to record a hologram than an in-line architecture. The high resolution of holographic film means this is acceptable in optical holography. However, in digital holography the bandwidth of the recording medium is already severely limited and if we are to extract information from reconstructions we need the highest possible resolution which, if one cannot harness the functionality of accurately reconstructing phase, is achieved through using an in-line architecture. Two of the most significant problems encountered with reconstructions of in-line digital holograms include the small depth-of-field of each reconstruction and corruptive influence of the unwanted twin-image. This small depth-of-field makes it difficult to accurately process the numerical reconstructions and it is in this shortcoming that we will make our first three contributions: focusing algorithms, background and object segmentation algorithms and algorithms to create a single image where all object regions are in focus. Using a combination of our focusing algorithms and our background segmentation algorithm, we will make our fourth contribution: a rapid twin-image reduction algorithm for in-line digital holography. We believe that our techniques would be applicable to all digital holographic objects, in particular its relevant to objects where phase unwrapping is not an option. We demonstrate the usefulness of the algorithms for a range of macroscopic objects with varying texture and contrast.