Photogrammetry #

• The flight plan for unmanned aerial vehicle is given below.
• The drone was flying at about 80 m above the ground.
• Photogrammetry is based on the overlap of the images. The same features at least has to be detected on 3 images.
• The images are then stitched. First step is densification and key-point detection.
• Once we have enough images of the overlapping images, the images are then calibrated and the camera position is estimated.
• The next step is to manually provide control points.
• The base-station was a RTK-based GNSS system (real time kinematic correction).
• The drone is the rover and the station is the base. The remote, the rover and the base are in communication all the time.
• This helps in the real time calculation of the geolocation and correction of the co-ordinates.
• The geolocation device is on the craft/rover as well as on the base station.
• For civilian usage, there is an error of 5 to 8 meters in the geo co-ordinates.
• It was connected to a CORS network (continuously operating reference station) and to a VRS (virtual reference station).
• The sensor was 1” CMOS 20 mega-pixel camera with 24mm lens and 84 deg field of view.
• The unmanned aerial vehicle had a automatic pitch-roll-yaw stabilization.
• The UAV flight had a flight altitude of 80m, flying at 7 m/hr, with 75% overlap and 80% side-lap.
• A total of 2,334 photographs were taken.
• The control points are added to the images manually and better control points give better measurements.
• The were occluded region during overall coarse scan. To fix that flat vertical scan profile photograph was taken with close by camera.
• The drone makes sure that is no motion blur while taking the photograph, by making sure the shutter speed, drone movement is condemn with each other.
• Since the photographs are taken in raw image format, all the parameters like gamma, color can all be corrected except the shutter speed.
• All this correction is all manual. There are some batch process, though, like tone, temperature and color balance so that there is no artifacts.
• Most of the batch processing is for tonal balancing. Tone balance can be done in batch process, but not gamma.

Rajarani camera details #

Terrestrial Laser Scan #

• The scanner was a FARO Focus M70 scanner.
• The scanner was operating based on time of flight laser pulse operating at 97 Hz scan speed at at about half a million points per second at a range of 1m to 30m with an accuracy of 3mm.
• A total of 62 scan position was fixed based on intricacy of the carvings on the temple and also to minimize the scan error.
• A built-in 360 deg 165 mega-pixel HDR camera with 2x exposure bracketing was used to capture color/texture information of the point cloud.
• The total time is 13 minutes for a given vantage point.
• Scanner was placed to minimize the coverage error.
• The scanner doesn’t have geo co-ordinates. The base station is required to tag geo co-ordinates.
• The of the geo co-ordinates in the point cloud merging happens in software.
• Some points that the scanner cannot reach will have to filled in manually from the photogrammetry data.
• The blind spots are hard to reach corners and top of the temple.
• When the data as been collected, the overlapping points are calibrated together in a process called registering.
• For point cloud registration is done manually by looking at the common overlapping points and triangulation of the points to get the absolute co-ordinates.
• The interior of the temple has no lights so therefore a torch was used at the garva gruha and the corridor connecting the garva gruha and the jagamohana.
• There was a metallic grill that was installed by ASI at the ceiling of the jagamohana and garva gruha.
• The grill was causing errors in registering the points. The laser scanner was placed at the top of the grill to take the scan of the ceiling of the jagamohana.
• There was a lot of bats between the grill and the ceiling of the garva gruha. The technician was scared to go in there. Some workers (gardeners I guess) had to shoo away the bats to let the technician to be able to put the scanner on the grill.
• Above the ceiling of the garva gruha there is an entrance to the hollow interior of the deula.
• The hollow interior of the temple was now nested by bats. A worker in the temple tried to take some photos of the interior. These are the photos he managed to take.