This forum is disabled, please visit https://forum.opencv.org
 | 1 | initial version |
I suggest two possible approaches:
1 - Manually crop the image (yes, it is time consuming, but at least is "accurate"). Google for some tools, like imageclipper. Or use gimp/related if is few images.
2 - To detect the eyes, you can use the eye classifier available in opencv/data/haarcascades. Or another method, but the eye cascade can help.
About the bad illumination conditions, try to apply some illumination normalization, for example, histogram equalization or gamma correction. The topic here have some suggestions and a reference to a paper.
| | 2 | No.2 Revision |
I suggest two possible approaches:
1 - Manually crop the image (yes, it is time consuming, but at least is "accurate"). Google for some tools, like imageclipper. imageclipper. Or use gimp/related if is few images.
2 - To detect the eyes, you can use the eye classifier available in opencv/data/haarcascades. Or another method, but the eye cascade can help.
About the bad illumination conditions, try to apply some illumination normalization, for example, histogram equalization or gamma correction. The topic here have some suggestions and a reference to a paper.
