MAIA multispectral camera Marco Dubbini Alma Mater University of Bologna Sec. Geography Mario Gattelli

Transcription

1 MAIA multispectral camera Marco Dubbini Alma Mater University of Bologna Sec. Geography Mario Gattelli

2 MAIA: The most advanced Multispectral Imaging Camera for drones (UAVs), aircraft and land-vehicles available today Photogrammetry-ready images in 8+3 VIS/NIR bands Bands are very close to WorldView2 satellite 2

3 MAIA at a glance Configuration and live image view over WiFi 8 bands cover the full VIS-NIR range + RGB Plug-and-play for immediate use Rich set of interfaces and full access to any camera setting for precise control 9x 1.2 Mpix sensor Global shutter 12 bits Factory lens geometric calibration Bundled scientific-grade image processing software Irradiance sensor with matched bands for reflectance measurements (patent pending - available soon) 3

4 Main applications Agriculture Precision farming Vegetative index mapping Water supply planning Optimize pesticide use Fertilization tuning Yield estimation Health condition monitoring Early detection of disease Environmental monitoring and survey Surface geology survey Spill of pollutant or hazardous substances Biomass mapping Industry Remote chemical imaging Large industrial plant monitoring 4

5 Architecture and interfaces MAIA USER interfaces Operator WiFi Ethernet Geo tagged images Video Out Radio Link Operator monitor Array of 9 synchronous sensors CPU Control signals Keypad Radio controller NMEA NMEA strings stings System interfaces Trigger in Gnss receiver Strobe Out Accelerometer Gyroscope Event Other Synchronized devices Gimbal Control Aerial or ground platform 5

6 Plugs I 6

7 Plugs II Nome Pin Signals Color Direction Logic Level Function J1 - GigEthernet Power IN Red Pwr 9-26V J2 2 Power GND Nero Gnd - Power J3 J4 J5 1 MPU SCL Giallo In TTL 3.3V 2 MPU SDA Blu In TTL 3.3V 3 GND Nero Gnd Red Reserved Bianco Reserved Verde Reserved Nero Reserved Verde Reserved Nero Reserved - 3 CH1 Blu In TTL 3.3V 4 CH2 Giallo In TTL 3.3V 5 RADIO GND Nero Gnd - IMU/GYRO - - RADIO 1 Serial TX Verde Out TTL 3.3V/RS Serial RX Blu In TTL 3.3V/RS-232 GNSS J6 3 Serial GND Nero Gnd - 4 Trigger Giallo In TTL 3.3V 5 Strobe Verde Out TTL 3.3V SYNC I/Os 6 GND Nero Gnd - J7 - HDMI

8 User interfaces Web interface Via WiFi before take-off On-screen display Via keypad before take-off Via Radio link during flight 8

9 Camera features Key technical specification 9 sensors: 1 RGB color + 8 bands in the VIS-NIR range ( nm). Bands are very close to WorldView2 satellite. 1.2Mpix CMOS sensor (1280 x 960) Global shutter Best-in-class sensitivity bits per pixel data depth Simultaneous shot of all sensors up to 6fps at 8 bit (3fps at bit) Geometric factory-calibration of each band. 3cm ground sampling distance (GSD) 45x34m 2 field of view (FOV) at 75m flight altitude above ground. Features and benefits Preliminary An extensive band set covering the full spectrum including 2 unique bands for pollution detection on water and soil + water supply planning. Sharp and deeply detailed images with no motion artifacts even with fast UAV speed. Data ready for precise data analysis and photogrammetric processing. Extremely detailed images with large area coverage for cost-effective data collection. continued 9

10 Camera features Key technical specification Full automatic or manual exposure mode, each sensor individually tunable. Trigger input and strobe output sync signals. 250/500 GB (Solid Disk) internal storage capability. Built-in WiFi and GigE interface with web-based interface for configuration, live image view and access to stored data. Input for external GNSS serial port with NMEA strings detection. Features and benefits Total control of the exposure parameters for high-quality images. Precise timing control of the image shot and sync of external devices. 5 hours of acquisition at 8bits, 1fps (1h at 3fps, 12bits). Preliminary Wireless on-field access via smartphone, tablet or PC for image stream and configuration. Fast cable data download at home. Accurate positioning of any image, for example with RTK GNSS continued 10

11 Camera features Key technical specification Remote controller input with event detection. Features and benefits Preliminary Easy control of key parameters during flight. Wide supply input (9-26VDC) and low power consumption (~7,5W). Compatible with all battery packages and limited impact on battery life. On-board 6-axis inertial unit with accelerometer and gyroscope (IMU). Small size (99x128x46)mm 3 Low weight (420g) Effective image stabilization and seamless integration into body. Eulerian angles writings in log file for each image. Fits fixed-wings and multi-rotor UAVs, with limited impact on other payload. 11

12 Ground Sample Distance DISTANCE (m) GSD (mm/px) FOV (mxm) Max Speed exp. time at 1ms (m/s) (km/h) Motion Blur < 0.50px x x x x x 157 >35 - >126 UAV AIRCRAFT 12

13 UAV 13

14 Aircraft 14

15 Spectral bands Band (nm) Approx. color Violet (Coastal) Blue Green Orange Red Red Edge NIR NIR2 9 (VIS) (RGB) The 8 bands of MAIA span the full VIS/NIR spectrum Bands are very close to WorldView2 satellite

16 Spectral bands Wavelength (nm) Coastal areas and marine environment Unaffected by atmospheric scattering Weak influence from atmosphere additional info on biomass Distinction between soil to vegetation Unaffected by atmospheric scattering Moisture content of biomass Reflectance peak for healty vegetation Plant classification Plant classification Yellowing of foliage Strongly depending on chlorophyll content Classification of vegetation Health status Absorbed by chlorophyll Detection of soil and geological features 16

17 Spectral bands Band width on the central wavelength Full width at half maximum Out of full width half band ±20 nm ±15 nm ±5 nm Indices coverage NDVI3, NG, NR, NNIR, RVI, GRVI, DVI, GDVI, SAVI, GSAVI, GOSAVI, MSAVI2, GMSAVI2 SR2, DI1, OSAVI, RDVI, MSR, MSAVI, MTVI, MCARI2, CAI, NPQI, SR15, NDVI4 Greenness index (G), SR1, SR4, SR6, NDVI, GNDVI, PSSRa, NDI1, PRI, SRPI, NPCI, PSNDa, PSNDc, PSSRc, SR11, SR18, NDVI8 SR7, PSSRb, NDI2, SIPI, HNDVI, MTCI, PSNDb, VIopt2, SR8, SR12, SR13, SR17, Viopt1, RGR, NDVI6, NDVI7 SR3, SR5, NDI3, MCARI, TCARI, TVI, CARI, ZTM, mnd705, msr705, SR9, SR10, SR14, SR16, DD, R-M, G-M, ND705, PSRI, NDIV1, NDVI2, NDVI5, REIP Considering the central wavelength of the eight bands, with a tolerance of ±5 nm 16% of the indices can be properly computed; increasing the tolerance at ±15 nm or ±20 nm the percentage raises respectively at 31% and 52% of the indices; considering the full width at half maximum, 72% of the indices can be computed 17

18 MAIA/S Spectral bands Wevlength (nm) Bandwidth (nm) Central Wavlength (nm) Bands are the same to Sentinel2 satellite (VIS/NIR)

19 Irradiance sensor 19

20 Irradiance sensor GNSS signal GNSS signal Irradiance sensor True reflectance measurement under any light condition for each bands independently Integrated cm-level GNSS positioning with Real Time Kinematics (L1 - RTK). Possibility to build wireless network of base stations for wide area coverage Patent pending RTK correction over LoRa radio link Base station Benefits Pixel-level geo-referenced and radiometric corrected images without ground targets (very timeconsuming task) To be released in late Q

21 Analysis flow 1/3 Images acquired by the camera Multispectral aerial survey of the area of interest 8+3 layers with several geometrical imperfections Image undistortion and co-registration! Factory-calibration geometric data + Parameters extracted from the images 8+3 pixel-to-pixel matching layers 21

22 Analysis flow 2/3 Digital numbers proportional to light hitting the target Reflectance values of the target Radiometric correction 3D model and geo-referenced orthophoto Performed by MAIA software Optional step, third-party software required 22

23 Analysis flow 3/3 A: Healthy vegetation, no actions required NDVI GNDVI SAVI B: Increase water supply by 10% Computation of relevant indexes (library of indexes included, possibility to define custom indexes) Analysis* and actions D: Check for potential disease in progress C: Start fertilization with XYZ Performed by MAIA software * Application-specialist involved (agronomist, geologist.) 23

24 Software GUI Powerful processing and unique features designed for users without background in photogrammetry. 24

25 Processing software features Key technical specification Image un-distortion and co-registration with factory-calibration data and context-sensitive analysis. Features and benefits Preliminary Perfect overlapping of the multi-bands images, with sub-pixel accuracy and undistorted images 3-modes radiometric correction for the bands: automatic, with reflectance target, or with irradiance sensor data. Automatic radial radiometric correction (vignetting). Index computation and false-color image generation with standard indexes of included library (NDVI, GNDVI, SAVI, ) and indexes defined by user at will. Preview of raw images and false-color index images. Accurate reflectance values for quantitative analysis. Ready-to-use data for widespread analysis and flexibility to explore new application-specific indexes. Immediate availability of relevant information. continued 25

26 Processing software features Key technical specification Flow-driven user interface with full parameter control and automatic processing option. Algorithms optimized for fast processing of very large datasets. Multi-layer and multi-band TIFF standard export formats. Features and benefits Preliminary Processing accessible to users without background in photogrammetry and to experts who needs scientific-grade analysis. Data from surveys of large areas quickly available to user, even on-site! Seamless integration with third-party remote sensing and photogrammetric processing software (ENVI, ERDAS, PCI Geomatics, Photoscan, Pix4D, ). Geometrical calibration parameters of each lens are available on request if user wants to process raw images autonomously 26

27 Conference and Paper International Conference on Agricultural Engineering. CIGR AgEng Aarhus, Denmark Agricultural Engineering International: CIGR Journal Last generation instrument for agriculture multispectral data collection Marco Dubbini a, Francesco Marinello b,*, Andrea Pezzuolo b, Michaela De Giglio c, Mario Gattelli d, Daniele Covi e a University of Bologna Sec. Geography, Bologna, Italy b Department of Land, Environment, Agriculture and Forestry, University of Padova, Legnaro, Padova, Italy c Department of Civil, Chemical, Environmental and Materials Engineering, University of Bologna, Bologna, Italy d SAL Engineering srl, Modena, Italy e Eoptis srl, Trento, Italy In press 27

28 MAIA multispectral camera Marco Dubbini Alma Mater University of Bologna Sec. Geography Mario Gattelli