Abstract
A self-tuning image restoration filter based on a simplified version of the Jaffe-McGlamery underwater image formation model is presented. Optimal values of the filter parameters are estimated automatically for each individual image by optimizing a quality criterion based on a global contrast measure. The simplified model is ideally suitable for diffuse-light imaging with limited backscatter, but qualitative tests show good performance in a variety of imaging conditions. In addition, quantitative tests with a large number of frames from six real mission videos indicate a substantial performance improvement when restoration is used as a preprocessor for a classifier detecting man-made objects in unconstrained subsea videos. © 2006 IEEE.
| Original language | English |
|---|---|
| Pages (from-to) | 511-519 |
| Number of pages | 9 |
| Journal | IEEE Journal of Oceanic Engineering |
| Volume | 31 |
| Issue number | 2 |
| DOIs | |
| Publication status | Published - Apr 2006 |
Keywords
- Computer vision
- Image restoration
- Self-tuning systems