Anurag Ohri, Manish Sahu Department of Civil Engineering, 3d Modeling of the Vishwanath Temple, Varanasi Using Close Range Digital Photogrammetry Anurag Ohri, Manish Sahu Department of Civil Engineering, Indian Institute of Technology (Banaras Hindu University) Varanasi, India 16th International Scientific and Technical Conference From Imagery to Map: Digital Photogrammetric Technologies November 14-17, 2016, Agra, India November 14, 2016
Introduction The Vishwanath Temple in Banaras Hindu University is the highly visited religious and tourist place of Varanasi. Well-known for their structural and surface complexity, temple constitute a great challenge to any attempt towards precise and detailed 3D measurements and modeling. This paper reports about a project using modern techniques of terrestrial and aerial photogrammetry to derive a photorealistic 3D model of Temple. The accuracy of model is improved by using control points collected using DGPS and Total Station. Accuracy of model is tested by using various GCP in the site.
Problems with Photogrammetric Survey in India In India Labor cost is very less. Monthly salary of Field Surveyor or CAD operator is $200-$700 per month. Average Size of Fields to be surveyed: 1 ha to 500 Ha Work done by small companies having very small capital. Photogrammetry is not very popular, only used in large projects Not trained manpower to operate UAV or do the processing of Aerial data. Myths that it is not possible to achieve high accuracy using Photogrammetry.
The Aim of Study To demonstrate the application of UAV photogrammetry in heritage mapping. To demonstrate the use of low cost UAV as a mapping tool for small areas. To do the accuracy assessment of map prepared by using low cost UAVs and software. To estimate the need of GCPs for achieving desirable accuracy.
About Shri Vishwanath Temple IIT(BHU) About Shri Vishwanath Temple Shri Vishwanath Temple is one of the most famous temples and biggest tourist attractions in the holy city of Varanasi. The temple is situated in Banaras Hindu University and is dedicated to the Lord Shiva. Shri Vishwanath Mandir has one of the tallest temple tower in India. Total height of the temple is Around 75 meters. Shri Vishwanath Mandir's construction took thirty-five years to complete (1931-1966). Total Area= 3.876 Hectare
UAV for Low Altitude Photogrammetric Mapping Small Unmanned Aerial Vehicles are very cheap. To perform very low altitude aerial photography Good for covering small areas. To get full image of buildings form different direction by flying at different heights and taking photographs at different angles To supply a cheap and easy system for engineering organization for high frequency needs of aerial photogrammetric survey
PLATFORM Vertical: 20 Camera Sensor: 1/2.3" Resolution: 12 MP FOV: 94° DJI Phantom 4 Quadcopter Location: GPS/GLONASS Location accuracy: Horizontal: 10 m Vertical: 20 Camera Sensor: 1/2.3" Resolution: 12 MP FOV: 94° Cost: US$1,500 only
Hardware & Software for Processing System: Desktop Computer Processor= Intel i7-4770, 3.4 GHz, 4 Cores RAM=12GB Hard Disk: 512 GB Graphics Card: 1 GB Operating System: Windows 7 , 64 bit Cost: US$ 1,500 Software for processing: Photomod UAS, Pix4D Mapper, Agisoft etc. Cost: US$ 375 per month to US$ 10,000 permanent License
Input Data Adding Images in the project with interior and exterior orientation parameter Camera calibration Parameters Block layout Creation Automatic Aerial Triangulation Automatic tie points Accepted Accuracy Measurement of GCP Block Adjustment Point Cloud Densification Geneartion of 3D textured Messh Generation of Digital Surface model Generation of Orthomosaic No No Yes Yes Workflow
Flight Planning Flying Height= 100 m above ground Level (Aerial Photograph) Both Aerial and Terrestrial Photographs Aerial Mission: Double Grid Aircraft Speed= 10 m/s Camera Angle=30° Oblique to Nadir Longitudinal Overlap=80% Side lap=50% Number of images= 335 Ground Resolution= 8 cm
Flight Plan
Image Properties
Ground control Point Instrument Used: Sokkia GRX2 GNSS Receivers and Topcon 5" accuracy Total Station. Number of GCPs: 6 Check Points: 13 Accuracy of GCPS: Horizontal: 2 cm, Vertical: 3 cm
Collection of GCP
GCP/MTP Manager
Location of 3D GCP and Check points
Camera Position
GCP and Automatic Tie Point
Average Ground Sampling Distance (GSD) 7.9 cm Camera Optimization 0.06% relative difference between initial and optimized internal camera parameters GCPs mean RMS error 0.026 m<GSD keypoints per image median of 12329
Computed Image and GCP Points Positions initial (blue dots) and computed (green dots)
Bundle Block Adjustment Details Number of 2D Keypoint Observations for Bundle Block Adjustment 1.8 million Number of 3D Points for Bundle Block Adjustment 0.66 million Mean Reprojection Error [pixels] 0.133 Internal Camera Parameters Focal Length Principal Point x Point y Initial Values 3.722 mm 3.159 mm 2.369 mm Optimized Values 3.724 mm 3.121 mm 2.274 mm Uncertainties (Sigma) 0.001 mm 0.002 mm
Ground Control Points
Check Points
Densify Point Cloud 3D Densify Poits:14 million Average Density (per cubic m)=19
Top view of Vishwanath Temple
3D Textured Mesh
3D Textured Mesh
Orthophoto
Digital Surface Model
Adding Grid Values to Point
Digital Elevation Model (IDW)
Time taken in different step Team Members: 2 Flight Planning: 30 minutes Flight to capture images: 1 Hour GCP Survey: 2 Hours Project initial Processing with GCPs: 2 hours 30 min Point Cloud Densification: 1 hour 10 min 3D Textured Mesh Generation: 40 min DSM and Orthomosiac Generation: 30 min Digitization of Point and DEM generation: 30 min Total: Approximate 4 man days
Problems with UAV based photogrammetry UAV technology not fully mature Need increase flight time, Obstacle avoidance, more regular paths, less crashes etc. Low quality of GPS/IMU sensors Safety Concern from UAV Privacy Concern Permission from Local Government due to flying restrictions Camera Problems Processing software not fully matured, need more research. Need of high speed cloud base processing. Software are very expensive, not affordable for small companies Training of manpower
Conclusion UAV is very cost effective and flexible device for capturing images from low altitude. More mature sensors in near future (GNSS, IMU, Cameras etc.) Many diverse applications. Much scope for data Processing methods. Accuracy is a concern, but suitable for many applications. In academics courses, need to include Photogrammetry. Legal and Security issues are main concern.