Abstract
A novel methodology is proposed to identify the distribution of the intensity of the illuminant across a real-world image; with this information, the image can be classified as a homogeneous or nonhomogeneous lighting source. A criterion is proposed for deciding whether the illumination in the tested image is homogeneous or nonhomogeneous. This criterion is based on well-established methods that are dedicated to improving color characteristics and are commonly used to determine the color of the lighting source during a captured scene. The classification of homogeneous or nonhomogeneous illumination in real-world images is implemented in the BSD500 database, and the results are compared with a manual classification of images—proposed under subjective criteria—by the authors; in this way, one can analyze how the lighting source affects the image. This procedure provides a classification methodology for real-world image databases found in the scientific literature.
Original language | English |
---|---|
Pages (from-to) | 825-836 |
Number of pages | 12 |
Journal | Color Research and Application |
Volume | 45 |
Issue number | 5 |
DOIs | |
State | Published - 1 Oct 2020 |
Keywords
- homogeneous
- illumination
- nonhomogeneous
- real-world images