Heritage conservators are interested in using digital images in the analysis of stained glass windows. Images of stained glass are significantly different from those of most other subjects because their colour is generated by transmitted, rather than reflected light. Also the medium has a wider dynamic range between highlight and shadow areas than most "real world" scenes. In many cases the background is partially visible through the glass, typically showing trees, foliage, sky or other buildings. Images of stained glass taken with external illumination very often contain shadows, moreover, cast by structures such as protective bars and grilles. The physical structures producing the shadows are often irremovable, because they are difficult to access or constitute structural elements of the window. It is thus necessary to provide a suitable set of image processing tools to remove both background and shadows from the digital images. This research was aimed at the problem of normalising a stained glass image (or a series of images) to remove artefacts arising from imaging geometry, illumination and background. The scope of the investigation covered the statistics of stained glass images, segmentation and feature extraction, and removal of image defects, including non-uniform illumination, shadows of bars and grilles, and background. This thesis introduces systematic procedures to calculate statistics of stained glass on digital images. Some interesting features are explained. The scale invariance property of images is examined carefully, and characteristic scaling behaviours of stained glass images are found. To developed novel algorithms for segmentation and feature extraction, and removal of image defects, including nonuniform illumination, shadows of bars and grilles, and background. These algorithms have been implemented using Mattab. These techniques enable the image defects and shadows to be characterised and removed with a reasonable degree of success.