1. Why Digital Cameras are Claimed to be Superior to Film Cameras
John Trinder
University of New South Wales
Sydney,
Australia

2. Statement on history of development of digital cameras
Topics discussed
Statement on history of development of digital cameras
Requirements for development of digital aerial cameras
Two solutions in development of digital aerial cameras
Pushbroom system
Leica Geosystems ADS40
Multiple frame cameras
Intergraph DMC
Vexcel UltraCamD
Characteristics of Digital Aerial Images
Reported Experiences with Digital Cameras
Advantages of digital cameras
Unresolved questions

3. Development of Digital Cameras
Steve Sasson (Kodak) developed the first digital camera 30 years ago
Based on small CCD sensors from Fairchild Corporation with 100x100 pixels.
Now years on, digital cameras have adequate resolution and size to match the quality of film cameras
Sasson states that film will still be used for photography for some niche areas
Digital imaging will largely replace film for almost all applications – [but how much for aerial photography?]
Future developments in photography are hard to predict, but almost limitless.
There are more uses for images and more images being taken than ever before.

4. High Resolution Digital Aerial Cameras
Digital photogrammetry has developed significantly over the past 15 years
Include efficient software for the production of DEMs, orthophotos and vector mapping
Development of digital aerial cameras has taken time, due to demands of:
Large area coverage
High spatial resolution – of the order of 400Mpixels or GSD of 10cm to 20cm
High geometric accuracy
Efficient management of TBytes of image data during imaging
It is expected that developments will take advantage of the characteristics of digital technologies

5. High Resolution Digital Aerial Cameras
Two solutions for development of digital aerial camera now available
Three linear arrays look forward, vertically and backwards to form three separate images as the aircraft moves over the terrain surface.
Images not perspective projections
System must include GPS/INS
Images from smaller area arrays are stitched together to form a larger frame image, which will have similar dimensions to a frame aerial film camera
Images will be perspective projections
No GPS/INS system required

6. Camera Concepts
ADS40
Single lens sensor with 10 channels generates endless pixel carpets
DMS
UltraCam
Multi lens sensors with up to eight lenses generates patchwork frames

7. Digital Camera Concepts
digital frames
pixel carpets

9. DATA ACQUISITION BY LEICA GEOSYSTEMS ADS40

10. composed of nadir view lines composed of forward view lines
Three-line Pushbroom Scanner
Backward scene
Nadir scene
Forward scene
composed
of backward view lines
composed of nadir view lines
composed of forward view lines
Backward
Nadir
Airborne Imaging Scenes:
The airborne digital sensor works like a push-broom scanner, recording continuously as the aircraft moves forward. When the recorded lines are put together they form a scene. Since three lines are recorded simultaneously three types of scenes are being formed:
the forward scene composed exclusively of forward view lines
the nadir scene composed exclusively of nadir view lines
the backward scene composed exclusively of backward view lines
Forward

11. Push-broom sensors
The geometry of the complete image is not a perspective projection. Hence, special software is required.
GPS/IMU system is essential to determine the camera exterior orientation (positions and attitude) during flight – extra cost
Linear arrays are less subject to loss of pixels
If bad pixels do occur, fewer pixels available to interpolate lost data
Linear arrays are claimed to have larger dynamic range
Linear arrays in principle are more suited to smaller scale imaging because of motion of the aircraft.
Linear array systems have recently demonstrated GSD of 5cm
Most linear array systems enable the acquisition of only 3 images per point along-track, but multiple imaging is possible across-track

12. ADS40 unrectified

13. ADS40 rectified

14. Multiple Frame Cameras
Intergraph (formerly Z/I Imaging) DMC
Vexcel UltraCam
Simultaneous acquisition of 4 frames - 7kx4k
Panchromatic - Syntopic data acquisition of 9 frames with camera at identical positions
Pixel size (pitch)
12 μm
9 μm
Field of View
69.3° cross track x 42° along track
B/H=0.30 for 60% o’lap
B/H=0.60 for 20% o’lap
55° cross track x 37° along track
B/H=0.27 for 60% o’lap
B/H=0.50 for 20% o’lap
Panchromatic images
Lens
f – 120mm/f4
f= 100mm/f5.6
Final output
7,680 x 13,824 pixels
7,500 (flight direction) x11,500 pixels

15. Multiple Frame Cameras Intergraph (formerly Z/I Imaging) DMC
Vexcel UltraCam
Multi-spectral images
Lens
f – 25mm
f-28mm/f4
Colours
B, G, R, IR, alternate IR
R, G, B, IR
Shutters and f-stop
1/50 – 1/300s, f4-f22
1/60 - 1/500s, f
Radiometric resolution
12-bit
Better than 12-bit
Ground sampling distance
500m
300m
FMC
yes

16. Multiple Frame Cameras
The images have the same perspective geometry as normal aerial images
No GPS/IMU system is required
There are many more neighbouring pixels from which to interpolate new pixel values for the erroneous data
Array imaging enables aerial triangulation of multiple redundant frame images leading to high geometric accuracy
If a high quality GPS/IMU system is installed for direct orientation, aerial triangulation may be avoided

18. Image Processing image mosaicing 4 overlapping images
apply camera calibration parameters
apply platform calibration
tie point area
tie point check
robust adjustment
projection to virtual perspective
fusion with colour composite

19. Combination of Sub-images to Create Single Virtual Image
3D-combination of sub-images
bundle adjustment of one sub-image to the other based on tie points
Generation of virtual image including geometric corrections  virtually image distortion free
sub-image
virtual image
Universität Hannover
Institut für Photogrammetrie und GeoInformation

20. Sample Image Test flight Germany Feb 2002 Flying Height = 150m
GSD = 1.5cm
Velocity ~ 140 kts (70 m/sec)
Exposure time = 1/100 sec
FMC Shift ~ 50 pixels

21. Sample Image
Checkered Tablecloth
Grid Size ~ 2 inches

22. Pan-Sharpening
Panchromatic Original Color Pan-sharpened
Color

23. Conclusions from Accuracy Tests
Geometric accuracy better than analog camera
Less favorable B/H ratio is compensated
Excellent height accuracy
Potential to fly at higher altitudes with fewer strips

24. Multiple area
arrays -
Vexcel
UltraCam

36. Characteristics of Digital Aerial Images
The cross-track coverage for pushbroom cameras is typical about 46º or less
Array cameras cross-track coverage is generally larger
Coverage does not approach that of standard film cameras.
Along-track coverage of area array cameras varies from 42º for the Intergraph DMC to 37º for Vexcel UltraCamD
B/H is about 0.7 B/H for push-broom cameras, 0.3 for are array cameras
With 80% overlap B/H is 0.6 for the DMC and 0.5 for the UltraCamD.
Improved quality of the digital images results in better quality image matching
Claimed to negate the impact of the smaller B/H.

37. Characteristics of Digital Aerial Images
Pixels sizes can vary from 5cm to 1m
The new generation digital aerial cameras will lead to a ‘paradigm shift’ in photogrammetry.
It should not be necessary to limit the number of photographs acquired on the basis of manual handling.
Highly redundant photography with much large percentage overlaps
Should result in significantly more reliable aerial triangulation of the images,
Image processing for elevation determination based on image matching from multiple images,
‘True’ or ‘near-true’ orthophotos will be acquired, in which the layover of buildings will be largely eliminated.

38. Reported Experiences with Digital Cameras
In Japan
ADS40 has been shown to have the potential for 1:2,500 scale mapping
Digital images will be applied to 1:1,000 mapping.
Map production is faster and more economical than with traditional photogrammetry.
Operations for mapping urban areas with very narrow streets with a Z/I Imaging DMS cost less than half that of film-based camera missions.
More than 12,000 images in 40 projects were collected in a period of 6 months.
Digital cameras can generate inexpensive true orthophotos, based on 80% overlaps along and across the strips.
The increased overlap allowed for the extraction of accurate DEMs for ground features as well as vertical structures.

39. Reported Experiences with Digital Cameras
In USA
Shortened production cycle and more accurate, realistic aerial images for tournament maps
ADS40 imagery acquired for orthophoto production over approximately 1 million square km - 10 Terrabytes of data in 3 month period
In Belgium
No significant difficulties in introducing the UltraCamD into operations and adapting to new system in flight planning.
More than 5000 images taken in three days and just a few days later the photos passed through quality assurance
Quoted accuracies of processing this data have been as high as 2-3 µm on the image
Image matching of digital images with larger overlaps more reliable than for standard aerial photography with 60% overlap.

40. Conclusions on development of digital cameras
Stated requirements:
Large area coverage
High spatial resolution – of the order of 400Mpixels or GSD of 10cm to 20cm
High geometric accuracy
Efficient management of TBytes of image data during imaging
Achievements:
Area of coverage is less than standard film cameras, but compensated by increasing overlaps
Eliminated degrading effects of film and improved dynamic range – resolutions less than 10cm achievable

41. Conclusions on benefits of digital cameras
Achievements (cont):
Despite lower B/H, achievable geometric accuracies are as good as or better, than for film aerial photogrammetry
Increased overlaps possible at little extra cost
Data management of large volume of data has been overcome
Other Benefits
Near ‘real’ orthophotos possible
More data acquisition per day and throughout the year, especially in higher latitudes
Processing of images can be done as soon as the aircraft lands, leading to more rapid throughput
New markets should be available for high resolution orthophotos and vector mapping
New opportunities in remote sensing with high resolution multi-spectral images

42. Unresolved Digital Camera Questions
Long term archival of digital data
Impact of calibration of multiple lenses
How robust and durable are the cameras?
What are the implications for calibration of cameras comprising up to 9 separate lenses?
What is the life span of digital cameras?
How quickly will they provide a return on investment?