Automatic Red-eye Detection and Correction Method

Jiajun BuJisuanji 2003

Abstract

Red-eye is the most common problem in image evaluation and correctio n. This paper presents a new efficient algorithm to detect and correct the red-e yes. The algorithm has been divided into five parts: (1) Get the improved HSI mo del using mixed color feature of the red-eyes; (2) Detect all the suspicious red regions automatically; (3) Use a filter to get the real red-eye regions automat ically; (4) Correct the red-eyes by reducing the color using the color features of the red-eye and the circumferences; (5) Gaussian smooth. To test the efficien cy, this algorithm has been realized using Visual C++ under Windows 2000 and the results show that it has fast processing speed and a good repair result.
BibTeX
@inproceedings{bu2003automatic,
  author = {Jiajun Bu},
  title = {{Automatic Red-eye Detection and Correction Method}},
  booktitle = {Jisuanji},
  year = {2003},
}

Loading related content...

Figure