|
|
EE 362 Final Project Color Balancing: The Battle of the Algorithms
Laura Diaz Jessica Heyman Gustav Rydbeck
|
|||||||||||||||||
| Introduction |
Test Images To test how well our algorithms performed, we decided to use a variety of pictures, all of which were generated using ISET with the camera settings shown in Table 1.
Table 1. We simulated nine different images under three different illuminants: tungsten, fluorescent and a flat power spectral distribution. The image with the flat power spectral distribution was obtained by selecting a white area in the scene, or if there was no suitable white in the image, the entire image was selected. The radiance of this area was then obtained and then divided out. The other two images were then created by imposing different illuminants on the scene by multiplying by the appropriate radiance distribution. The three resulting images, one for each illuminant, were then saved as three dimensional matrices containing the RGB values for each pixel. The images we chose represent both indoor and outdoor scenes. Some images contained a wide range of colors, while other had many specular objects or contained a lot of contrast. Examples of the images created are shown in Figure 1.
Top : Input Image, Gray World (GW), White World (WW), Scale By Max (SBM) Middle: Mean/Std (MS), GW/Std, Standard Deviation (Std), SBM/MS, Bottom: MS/SBM, WW/MS, MS?WW, WW/Std Figure 1. Sample images used for algorithms
|
