Outline

1
(No Transcript)
2
Outline

• About Surdex
• Digital Acquisition Trends
• Image Processing
• Image Processing Approach
• Questions and Answers

3
Surdex

• Surdex headquartered in St. Louis, MO area
• Over a half-century of photogrammetric mapping
• 100 employees
• 7 aircraft 4 Cessna 441 (Conquest) Turbine
  Aircraft
• 3 Intergraph Digital Mapping Cameras (DMC)
• 5 film cameras
• LiDAR
• Very progressive RD
• Large amount of proprietary algorithms and
  software
• Focus on cost-efficiency and performance

4
Surdexs Digital Camera

• Own and operate 3 Intergraph DMCs
• Digital Mapping Camera
• Operated since summer, 2005
• Large-format digital frame camera
• Simultaneous capture of panchromatic, color, near
  infrared
• 12 bits/pixel potential dynamic range
• 500MB per exposure
• gt600,000 exposures processed

5
Multispectral Imagery (Minnesota NAIP, 2008)
6
Multispectral High-Resolution Imagery
7
3 Resolution (Ottawa, Canada, 2007)
8
Digital Imaging From 4 Miles Up
2 DOQQ Close-Up
9
Digital Acquisition Trends

• Film ? digital faster than most experts
  imagined
• 40 large-format digital cameras in North America
• Doubled in last 3 years
• Numerous companies have completely abandoned film
• Film suppliers reducing emulsion offerings
• Trends
• Increasing demand for 4-band (R,G,B,NIR)
• Increasing demand for stereo
• Increasingly aggressive schedules
• Many operators have developed their own software

10
Image Processing for Results

• What does the client want?
• Pretty picture?
• Support detailed GIS analysis?
• Match the last project?
• Hardcopy or softcopy?
• Client must be involved in establishing
  colorimetry
• Color easiest to establish
• Contrast affects brightness and sharpness
• Brightness interlocked with contrast
• Digital imagery what color do you want?

11
Resolutions Flying Height
Actual Surdex DMC Test Imagery All Exposures
Taken Within 1 Hour Timeframe All Remapped to
(target) 0.5 GSD
0.4 (4,000) 0.5 (5,000) 0.6 (6,000) 0.7
(7,000) 0.8 (8,000) 20 low nominal 20
high 40 high 60 high
12
Color Variation
Courtesy ITT Space Systems
13
Calibrated Monitors are a Must!
Actual Pictures of Same Image on Different
Monitors!
Courtesy ITT Space Systems
14
Linear CCD response needs correction
Gamma (Logarithmic) Correction
15
Processing of Digital Imagery

• Use reference images from acquisition
• Work with client to get the desired look
• Lock down colorimetry before production begins
• Why?
• Client has ownership in what deliverables look
  like
• Less alteration of the imagery in the last step
• Less re-work, better performance (schedule)
• Reduces QC to accuracy and artifact scrutiny

16
Key Points

• Technicians sort raw images into groups
• Geopositioned using ABGPS/IMU
• Common scene characteristics
• 1,000s of images loaded at once
• Enhancements are prototyped
• Batch-processing in distributed environment
• 12bpp processing option to output to 8bpp
• 4-band processing to output color and CIR

17
Processing

• Gamma correction
• Initial targeting of image metrics
• Digital dodging
• Offset atmospheric haze
• Offset localized brightness/contrast variations
• Solar correction (hot spot removal)
• Offset solar illumination (especially for
  high-altitude)
• Aided by sun ephemeris data and ABGPS
• Automated with operator override
• Full 12 bpp processing
• Option to output in 8 bpp at completion

18
Image Processing Flow Example (NAIP08)
1. Grouping of images 2. Gamma correction 3.
Correct to color metrics
Dodging to remove local variations
Solar correction to offset sun illumination
19
Image Processing Flow
Uncorrected 12-bit Raw Image
20
Image Processing Flow
Initial Gamma Correction
21
Image Processing Flow
Solar Correction
22
Image Processing Flow
Digital Dodging
23
Pre-Production vs Final Product WYSIWYG
24
Effectiveness of the Approach

• NAIP 2009
• 6 states -- 540,000 sq mi
• 100,000 exposures acquired _at_ 34,000 MSL
• 38,000 DOQQs _at_ 1m resolution X 4-bands
• 5.5 months from acquisition through final
  delivery
• Despite incremental product processing
• Achieve overall state consistency
• Achieve national consistency
• The keys
• Reference imagery process
• Colorimetry established before production

25
(No Transcript)
26
(No Transcript)
27
Delivering Seamline Data

• New deliverable to all clients
• Seamline polygons with attributes
• Date/time of acquisition
• Camera aircraft IDs
• Size of polygon
• Why?
• Provide guidance to client or 3rd party QC most
  errors are related to seamlines
• Provide a date for each pixel desired by some
  for historical reasons

28
Common Seamline Error and Resolution
29
Seamlines Spokane, WA 2009
30
Closing Comments

• Clients and contractors must work together to
  establish colorimetry prior to production
• No surprises upon delivery
• Assures schedule
• Reduces cost
• Take advantage of the content of digital sensors
• 12bpp 8X more data than scanned film
• 4-band (color CIR)