Underwater Imaging and Optics: Recent Advances Frank M. Caimi 1 , Donna M. Kocak 2 , Fraser Dalgleish 3 , John Watson 4 1 P.O. Box 700367 Wabasso, FL 32970 2 Maritime Communication Services / HARRIS Corporation 1025 West NASA Blvd. Melbourne, FL 32919 USA 3 Harbor Branch Oceanographic Institution at FAU 5600 N US 1 Fort Pierce, FL 34946 USA 4 University of Aberdeen King’s College Aberdeen AB24 3FX UK Abstract – Obtaining satisfactory visibility of undersea objects has been historically difficult due to the absorptive and scattering properties of seawater. Mitigating these effects has been a long term research focus, but recent advancements in hardware, software, and algorithmic methods have led to noticeable improvement in system operational range. This paper is intended to provide a summary of recently reported research in the area of Underwater Optics and Vision and briefly covers advances in the following areas: 1) Image formation and image processing methods; 2) Extended range imaging techniques; 3) Imaging using spatial coherency (e.g. holography); and 4) Multiple-dimensional image acquisition and image processing. I. INTRODUCTION Recent advancements in the field of Ocean Optics are, at least, partially attributable to the following developments: • Affordable, high quality cameras that support a suite of fast, inexpensive specialized image processing software and hardware add-ons; • Digital holographic cameras that record interference fringes directly onto solid state sensors (i.e., mega pixel charge coupled devices) to produce time resolved, 3-D movies; • High repetition rate, moderate power lasers and advanced detector designs which enhance performance of two- dimensional (2-D) and three-dimensional (3-D) imaging systems; • Compact, efficient and easy to program digital signal processors that can execute algorithms once too computationally expensive for real-time applications; • Modeling and simulation programs that more accurately predict the effects that physical ocean parameters have on the performance of imaging systems under different geometric configurations; • Image processing algorithms that handle data from multiple synchronous sources and that can extract and match feature points from each such source derive accurate 3-D scene information; and • Digital compression schemes provide high-quality standardizations for increased data transfer rates (i.e. streaming video) and reduced storage requirements. The applicability of each of these is unique to general topical areas in undersea imaging covered in the following sections. A more thorough discussion of these topics is found in [1]. II. CONVENTIONAL IMAGE ACQUISITION FOR SURVEY AND MONITORING Many recent applications are aimed at deploying conventional cameras for long-term monitoring and observation where collection of video imagery and subsequent processing can be used to provide image scaling and measurements, identification and assessment, 3-D reconstructions, and analysis [2-5]. These systems benefit from cameras capable of operating continuously for long, unattended durations with ample resolution and high bandwidth data links. One recent example is a high-speed megapixel benthic imaging system designed to operate from a towed platform for scallop stock assessments [6]. The system features a commercial GigE Vision TM camera (1360 x 1024 pixels), strobe light, and tow vessel that streams images at 4/s for near real-time monitoring. Tow speeds of 5 – 8.5 km/h provide 58 – 27% overlap between successive images with a resolution of greater than 1 pixel per mm, enabling identification of objects less than 50 mm in diameter in most water conditions. The system is also useful for fine-scale habitat mapping, ground-truthing acoustic data, benthic ecology research and fishing damage assessment. Higher accuracy imaging equates to increased pixel resolution requiring much greater bandwidth and storage capability. High definition (HD) and ultra high definition video (UHD or super hi-vision) produce images having 1920 x 1080 and 7,680 x 4,320 pixels, respectively. Image compression formats such as HDV, MPEG-4 AVC/H.264 and VC-1 reduce data rate and memory demands by up to a factor of nearly 50 [7, 8]. Although these reductions help make requirements more manageable, users are often not in favor of “losing” data. Table 1 provides approximate upper bound bandwidth requirements for various types of cameras, including some with image compression. 978-1-4244-2620-1/08/$25.00 ©2008 IEEE