1 Elizabeth Wirsching1, Ramesh Sivanpillai2, 3 & Greg Brown2Quantitative Rating System for Imagery Acquired with Unmanned Aerial Vehicle Elizabeth Wirsching1, Ramesh Sivanpillai2, 3 & Greg Brown2 1. Department of Ecosystem Science & Management, 2. Department of Botany, 3. Wyoming GIS Center University of Wyoming
2 My Background 6 years in the United States Air ForceImagery intelligence (geospatial analyst) U-2 mission planner
3 Remote sensed imagery applicationsMilitary/Reconnaissance Agriculture Industry Urban/City planning
4 UAVs or drones Flexible Affordable Images and videos Different sensors
5 Platforms Space-born Landsat, Weather satellites (GOES)High/Medium-altitude U2, Military Predator Low-altitude Aircraft, Drones
6 UAV or drone images Stability Crabbing (overlap) Drifting(Pitch, roll, and yaw) Crabbing (overlap) Drifting
7 Image distortion (1972 U2 and Landsat 1 images)
8 Problems with UAV imageryDistortion or other problems (poor contrast, too bright, fuzzy or sharpness,) can diminish the use of these images UAVs are very flexible Chances of distortion and other problems could be higher Before delivering the images they have to be rated for their usefulness – so users can select the appropriate ones
9 National Image Interpretability Rating ScalesNIIRS – National Image Interpretability Rating Scales Military Civilian Natural Cultural/Urban/Industrial Agricultural
10 NIIRS 0 No value NIIRS 1 Distinguish between major land use classes (e.g., urban, agricultural, forest) Detect a medium-sized port facility. Distinguish between runways and taxiways at a large airfield. Identify large area drainage patterns by type (e.g., dendritic, trellis, radial). NIIRS 2 Identify large (i.e., greater than 160 acre) center-pivot irrigated fields during the growing season. Detect large buildings (e.g., hospitals, factories). Identify road patterns, like clover leafs, on major highway systems. Detect ice-breaker tracks. Detect the wake from a large (e.g., greater than 300') ship NIIRS 3 Detect large area (i.e., larger than 160 acres) contour plowing. Detect individual houses in residential neighborhoods. Detect trains or strings of standard rolling stock on railroad tracks (not individual cars). Distinguish between natural forest stands and orchards NIIRS 4 Identify farm buildings as barns, silos, or residences. Count unoccupied railroad tracks along right-of-way or in a railroad yard. Detect basketball court, tennis court, volleyball court in urban areas. NIIRS 5 Identify Christmas tree plantations. Identify individual rail cars by type and locomotives by type. Detect open bay doors of vehicle storage buildings. Identify tents (larger than two person) at established recreational camping areas. NIIRS 5 cont. Distinguish between stands of coniferous and deciduous trees during leaf-off condition. Detect large animals (e.g., elephants, rhinoceros, giraffes) in grasslands. NIIRS 6 Detect narcotics intercropping based on texture. Distinguish between row (e.g., corn, soybean) crops and small grain (e.g., wheat, oats) crops. Identify automobiles as sedans or station wagons. NIIRS 7 Identify individual mature cotton plants in a known cotton field. Identify individual railroad ties. Detect individual steps on a stairway. Detect stumps and rocks in forest clearings and meadows NIIRS 8 Count individual baby pigs. Identify a USGS benchmark set in a paved surface. Identify grill detailing and/or the license plate on a passenger/truck type vehicle. Identify individual pine seedlings. Identify individual water lilies on a pond. Identify windshield wipers on a vehicle. NIIRS 9 Identify individual grain heads on small grain (e.g., wheat, oats, barley). Identify individual barbs on a barbed wire fence. Detect individual spikes in railroad ties. Identify individual bunches of pine needles. Identify an ear tag on large game animals (e.g., deer, elk, moose)
11 NIIRS 0 No value NIIRS 1 Distinguish between major land use classes (e.g., urban, agricultural, forest) Detect a medium-sized port facility. Distinguish between runways and taxiways at a large airfield. Identify large area drainage patterns by type (e.g., dendritic, trellis, radial). NIIRS 2 Identify large (i.e., greater than 160 acre) center-pivot irrigated fields during the growing season. Detect large buildings (e.g., hospitals, factories). Identify road patterns, like clover leafs, on major highway systems. Detect ice-breaker tracks. Detect the wake from a large (e.g., greater than 300') ship NIIRS 3 Detect large area (i.e., larger than 160 acres) contour plowing. Detect individual houses in residential neighborhoods. Detect trains or strings of standard rolling stock on railroad tracks (not individual cars). Distinguish between natural forest stands and orchards NIIRS 4 Identify farm buildings as barns, silos, or residences. Count unoccupied railroad tracks along right-of-way or in a railroad yard. Detect basketball court, tennis court, volleyball court in urban areas. NIIRS 5 Identify Christmas tree plantations. Identify individual rail cars by type and locomotives by type. Detect open bay doors of vehicle storage buildings. Identify tents (larger than two person) at established recreational camping areas. NIIRS 5 cont. Distinguish between stands of coniferous and deciduous trees during leaf-off condition. Detect large animals (e.g., elephants, rhinoceros, giraffes) in grasslands. NIIRS 6 Detect narcotics intercropping based on texture. Distinguish between row (e.g., corn, soybean) crops and small grain (e.g., wheat, oats) crops. Identify automobiles as sedans or station wagons. NIIRS 7 Identify individual mature cotton plants in a known cotton field. Identify individual railroad ties. Detect individual steps on a stairway. Detect stumps and rocks in forest clearings and meadows NIIRS 8 Count individual baby pigs. Identify a USGS benchmark set in a paved surface. Identify grill detailing and/or the license plate on a passenger/truck type vehicle. Identify individual pine seedlings. Identify individual water lilies on a pond. Identify windshield wipers on a vehicle. NIIRS 9 Identify individual grain heads on small grain (e.g., wheat, oats, barley). Identify individual barbs on a barbed wire fence. Detect individual spikes in railroad ties. Identify individual bunches of pine needles. Identify an ear tag on large game animals (e.g., deer, elk, moose)
12 UW Project (2017) Goal is to observe epiphytes in tropical forests with little disturbance Pilot study in UW to test the capabilities of UAV technology Four small plants were strategically placed in or around trees in a closed space Images of these plants were acquired with an UAV https://commons.wikimedia.org/wiki/File:Epiphytes_perched_on_kauri_branch.jpg
13 UW Project (2017) DJI Phantom 2 Vision+14 Megapixel camera True color images Williams Conservatory in UW campus Photos acquired by DJI were rated with the adapted the NIIRS system Dji.com
14 Modified NIIRS Retained 0, and 6-9 from NIIRSRenumbered them as 0, 1, 2, 3, and 4 Added a new category (5) Able to distinguish between male and female flowers Identify number of petals on flower Detect male and female flowering parts
15 Quality rating is based on:Overall rating of visibility of the target Ranges between 0 (poor) to 5 (excellent) Five additional criteria Sharpness (focus) Noise/Artifacts 0 (lot of noise) to 5 (no noise) Contrast Brightness Distortion
16 VTY_SNCBD_MMDDYYYY UAV Image Rating Number (UIRN)Every photo is rated by an interpreter First 3 digits – overall value and type Next 5 digits – individual image quality ratings on 5 criteria Last 8 digits – image acquisition date (MMDDYYYY) VTY_SNCBD_MMDDYYYY
17 3TC_45534_04252017 UAV Image Rating Number (UIRN)Over all value rating based on the target (0 – 5)
18 Sensor and/or image type in this case True ColorUAV Image Rating Number (UIRN) 3TC_45534_ Sensor and/or image type in this case True Color
19 3TC_45534_04252017 UAV Image Rating Number (UIRN)Sharpness value 0 (poor) – 5 (excellent)
20 UAV Image Rating Number (UIRN)3TC_45534_ Noise (0-5)
21 UAV Image Rating Number (UIRN)3TC_45534_ Contrast (0-5)
22 UAV Image Rating Number (UIRN)3TC_45534_ Brightness (0-5)
23 UAV Image Rating Number (UIRN)3TC_45534_ Distortion (0-5)
24 UAV Image Rating Number (UIRN)3TC_45534_ Date (MMDDYYYY)
25
26 5TC_34545_
27 0TC_55555_
28 1TC_45555_
29
30
31
32 Notes First digit (Value) applies for RGB or true color sensors onlyFor other sensors (IR, RADAR) new value scale has to be applied
33 Benefits Users can efficiently search for certain types of imagesExamples: high contrast images only, high sharpness and brightness With all parts of the rating system having the same range (0-5) there is less confusion
34 Future work Level of detail necessary to identify the speciesOther conditions in which epiphytes Newer models (smaller drones) Off-the-shelf models versus build-your-own Fly outdoors
35 Thank you Nic White (Pilot) Elizabeth Traver (WyCEHG Manager)WyCEHG (Phantom DJ Vision 2) WRSP (funding support)
36 Questions?