Deep Underwater Surveys by Photogrammetry for
Transcription
Deep Underwater Surveys by Photogrammetry for
Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Pierre Drap CNRS, LSIS laboratory Marseille, France Jean-Christophe Sourisseau, Archaeologist, CCJ, Aix en Provence, France Timmy Gambin, Archaeologist, University of Malta Luc Long Archaeologist DRASSM, France Joaquim Garraboum CSIC, Marine Biology, Barcelone, Spain Bertran Chemisky, COMEX, Marseille, France Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 o What should be a survey ? Filling the gap between knowledge and geometry … o Underwater photogrammetry for archaeology, focusing on geometry. The beginning, close to Marseille: 1970 o Deep Underwater Survey by photogrammetry: The Geometry o The photogrammetric tool o Application to underwater archaeology o Application to marine biology o What about knowledge? o Ontologies for photogrammetry and archaeological artifact o Using knowledge for 3D reconstruction in naval archaeology o Merging survey with different resolution Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 What should be a survey ? - A survey is an interpretation of the «real world» - need to be connected with documentation - need a measuring tool • • • • • • Hand made design Photography Photogrammetry Laser scanner SONAR … -… 3D Scan mensi Andrea Palladio Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology 1970 First underwater photogrammetric survey for archaeology in Marseille, Bernard Lioux, Wreck 3, Planier, Marseille. Sorrento, 16th April 2015 Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Seven years before, 1963, Stereo photogrammetry from submarine Yassi Ada, 1963, 35 meter deph Submarine Asherah Georges Bass Sorrento, 16th April 2015 Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology La Madrague de Giens, 1972 Giens, France Antoinette Hesnard Photogrammetry and Cartography according to specific request Done in 2008 Sorrento, 16th April 2015 Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 Wreck: Cap Bénat 4, 328 m 1981, Cyana Ifremer DRASSM By courtesy Luc Long Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 Cap Bénat, 328m, 1981 Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Plage d’Arles 4, 662 m 1993 Sorrento, 16th April 2015 Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 Plage d’Arles 4, 662 m Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 Plage d’Arles 4, 662 m Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 Plage d’Arles 4, 662 m 1993 Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 Grand Ribaud F (61 m) DRASSM 2000-2002 COMEX CNRS Photo: F. Bassemayousse © Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Grand Ribaud F (61 m) Sorrento, 16th April 2015 Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Grand Ribaud F (61 m) Deep excavation by blaster Sorrento, 16th April 2015 Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 3D Reconstruction Photograph Photo: F. Bassemayousse Grand Ribaud F (61 m) Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 Grand Ribaud F (61 m) Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 The rov3D project, COMEX - CNRS/LSIS Constraints: • • • • • • Non intrusive survey, no contact High accuracy result Large scale result Possibly embedded in a small ROV No sophisticated device, inertial central, High resolution acoustic device, ect… Real time evaluation in order to be sure all needed data are captured s Proposed solution • • • • • • • Full optical survey, photogrammetry Three digital camera synchronized Two low resolution, one high resolution One computer embedded in the ROV for storage and preliminary process One computer on board, recieving data from ROV and building Visual odometry real time with ROV trajectory and sparse cloud of 3D points Off line, days after, high resolution 3D model reconstruction Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Synchro bloc 10Hz, 2Hz 2 Hz 10 Hz 10 Hz Sorrento, 16th April 2015 Synchronized LED strobe Calibrated triplet camera bar Image memory management 10 Hz Extraction feature Points from 2 stéréopair at t, t-1 Computer Embedded in ROV rov cable, IP com 3D point computation at 10 Hz 3D / 2D Visualization 3D + coulour Visual odometry ≈ 300 pt Surface computer used for ROV guidance, real time ROV location display Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology rov 3D / real time survey Devices on ROV - 3 synchronized camera 1 High resolution (6576x4384) 2 Low resolution (1936x1456) - 1 computer embedded in ROV - Ethernet com Sorrento, 16th April 2015 Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 rov 3D: Experimention on Cap Bénat wreck, 328m depth First original survey, 1981, now from ROV: Visual odometry Real time Guidance help Short results given on board Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 rov 3D: Experimention on Cap Bénat wreck, 328m depth 3 hours on site, ROV, 2 days process off line. Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 rov 3D: Experimention on Xlendi wreck, Malta, 100 m depth From submarine, both visual inspection and survey Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 rov 3D: Experimention on Xlendi wreck, Malta, 100 m depth 3 hours on site, ROV, 2 days process off line. Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 rov 3D: Experimention on Xlendi wreck, Malta, 100 m depth Knowledge representation: ontologies on photogrammetry and CH connection (alignment) between local CH object and CIDOC‐CRM OWL / JAVA Procedural attachment CIDOC CRM Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Founding known object in 3D cloud of points Using 3D Descriptor extraction for interest point A lot of methods Chua et Jarvis “Point Signatures” Sun et Abidi 3D “Point Fingerprint” Johnson “Spin images” Tombari “SHOT” We develop another method: 2D slicing or ‘level curve’ - Sorrento, 16th April 2015 Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 Founding known object in 3D cloud of points Founding an instance of a theoretical model present in a measured site … with several problems, resolution accuracy … Slice method is fine but … need approximate values … Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 Ships were built by people, and human error, combined with the limitations of the building materials, make the final result different from the conceived model: a) b) c) d) José Luis Casaban with Filipe CASTRO, lines are not parallel, surfaces are not even, distances are not regular edges are eroded... University of TEXAS, TAMU Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 Red coral measurment Often by diver because Dificult to reach Need to choose the right part Difficult to take oly two good images Orient 2 to 10 Use by stereo pair Impossible to use correlation The tree structure is an obstacle to automation Manual shooting Manual measurement Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 Managing data two kind of data: ‐ Random quadrat ‐ Fixed transect ‐ We have more than 10 years of photogrammetric campaign ‐ Difficulties: ‐ Having a continuity, In the method, people, software Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 Photogrammétrie Genetic study High definition Cartography Genetical distance High definition survey of coral structure Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 White coral measured in canyon de Cassidaigne, close to Marseille, France. 250 m depth. Challenge: complex terrain, non DEM representation… 33 Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015 Deep Underwater Surveys by Photogrammetry for Underwater Archeology and Marine Biology Sorrento, 16th April 2015