使用攝像設備架設於水面上對水下地形、地物進行拍攝時，受限於水面幾何因水流與風場的影響而產生的動態變化，以及陽光在水面上造成的波光影響，不易取得形狀完整以及幾何正確的影像。水下攝影設備可應用於水深較大的地區，但在水深小於 40 公分的淺水區域，水下攝影設備受限於作業空間，無法在此水深範圍內使用。
本研究使用水面上定點連續拍攝取得影像資料，輔以眾數濾波器以及平均濾波器得到水面水平時的像元灰階資訊，並校正拍攝期間陽光強度變化造成影像亮度不同的影響，得到幾何正確的水下影像。本研究之實驗地點包含美濃黃蝶翠谷與台北縣烏來鄉的桶后溪流段，依流況分區拍攝。本研究並分析眾數濾波以及平均濾波兩種方法所產生影像的優劣，利用MTF曲線評估兩者影像的解析力，及估計個別方法得到完整影像成果所需要拍攝的張數。
本研究實驗成果顯示，對於流況較緩之區域，使用本研究所提之方法可獲得清晰的水下影像，而流況較劇烈之區域所得的成果雜訊較多，但仍可恢復正確的水下影像幾何。而眾數濾波器所得的影像成果其解析力優於平均濾波器之影像成果，但需較多的拍攝時間及張數。

Underwater features at shallow depth (~40cm) are difficult to be observed and monitored. The water surface disturbed by flow and wind action prevents the direct observation from above water. Water-proofed camera or video are also hard to maneuver in such situations.
In this research, we tackle this problem by using continuous photos (1500~3000 photos per site) taken from above water and setup on a tripod. We employed a Nikon D200 camera, which takes about 12~24 minutes for each site. A grey card of photogrammetric grade is used to adjust the temporal change of illumination condition. Test sites include several flow types, i.e., no perceptible flow (NP), smooth (SM), unbroken/broken standing waves (USW/BSW). Mode-based filter and mean-based filter are developed to filter the pixels that are disturbed by wave actions. We analyzed the image resolution of the filtering results by using MTF curves, and estimated the number of photos needed to taken for each filter.
The results show that the proposed filters are most effective for the flow condition associated with NP and SM. For more dynamic flow condition, such as that associated with USW and BSW, a clear image is not obtained, but the geometry is preserved. The mode-based filter approach is needed more photos to taken than mean-based filter, but the image resolution is better.

經濟部水利署
Http://www/wra.gov.tw， 2010年4月28日

Google map
http://maps.google.com.tw/，2010年5月15日

Schneider Optics
http://www.schneideroptics.com/， 2010年5月8日

Scottish Environment Protection Agency
http://www.sepa.org.uk/， 2010年5月1日

Sedimetrics® Digital Gravelometer™ documentation
http://www.sedimetrics.com/documentation/introduction.html，2010年7月17日

陳維正，「應用相對基本法之立體影像修正」，國立雲林科技大學電機工程研究所碩士學位論文，2004。

Butler, J. B., Lane, S. N., and Chandler, J. H., “Through-Water Close Range Digital Photogrammetry in Flume and Field Environments,” Photogrammetric Record, Vol.17, p.419-439, 2002.

Graham, D. J., Reid, I., and Rice, S. P., ” Automated Sizing of Coarse-Grained Sediments: Image-Processing Procedures,” Mathematical Geology, Vol.37, No.1, 2005.

Graham, D. J., Reid, I., and Rice, S. P., ” A transferable method for the automated grain sizing of river gravels,” WATER RESOURCES RESEARCH, Vol.41, 2005.

Hecht, E., Optics. 3rd edition, Addison Wesley, 1998.

Mobley, C. D., Light and Water. Academic, New York, (1994).

Reid, M. A., Thoms, M. C., “Surface Flow Types, Near-Bed Hydraulics and The Distribution of Stream Macroinvertebrates,” Biogeosciences, Vol.5, p.1043-1055, 2008.

Shaw, J. A., Churnside, J. H., “Scanning-Laser Glint Measurements of Sea-Surface Slope Statistics,” APPLIED OPTICS, Vol.36, NO.18, p.4202-4213, 1997.

Whitman, M. W., Moran, E. H., and Ourso, R. T., ”Photographic Techniques for Characterizing Streambed Particle Sizes,” Transactions of the American Fisheries Society, Vol.132, p.605-610, 2003.