This chapter reviews basic statistical properties of photographic images, as observed empirically, and describes several models that have been developed to incorporate these properties. The models discussed are the Gaussian model, wavelet marginal models, and wavelet joint models. These models have been validated by examining how well they fit the data, but the true test usually comes when one uses the model to solve an image-processing problem. Some simple denoising examples have been presented to give an indication of how much performance gain one can obtain by using a better statistical model. The discussion is limited to discretized gray-scale photographic images. Many of the principles are easily extended to color photographs, temporal image sequences (movies), and more specialized image classes such as portraits, landscapes, and textures. In addition, the general approach may also be applied to nonvisual imaging devices, such as medical images, infrared images, radar and other types of range images, and astronomic images.
|Original language||English (US)|
|Title of host publication||Handbook of Image and Video Processing|
|Number of pages||11|
|State||Published - 2005|
ASJC Scopus subject areas
- Computer Science(all)