Airborne:
Color Infrared Digital Camera System Used for Natural
Resources Surveys
By Thomas J. Bobbe and Joseph P. Zigadlo
Introduction
Color infrared (CIR) aerial
photography is widely used for a variety of natural
resource interpretation and mapping applications. Although
CIR film was developed primarily for camouflage detection
during World War II, the combination of green, red and
near infrared spectral information has proven to be
valuable for distinguishing subtle differences in
vegetation species and condition. A new development in
electronic imaging has provided another method for
collecting CIR imagery. Recently the USDA Forest Service,
Nationwide Forestry Applications Program evaluated a CIR
digital camera system for remote sensing applications.
Digital camera systems are
an emerging electronic imaging technology that has
excellent potential for natural resource remote sensing
applications. Digital cameras can provide high quality
imagery in near real-time to complement the use of other
remote sensing imagery such as conventional aerial
photography, airborne video and satellite imagery to
collect resource information. Currently several
panchromatic and natural color digital camera systems are
commercially available. The Nationwide Forestry
Applications Program has evaluated the Kodak DCS 200 and
DCS 420 natural color digital cameras for resource
interpretation and GIS data collection. Based on the
success of these preliminary tests, the Nationwide
Forestry Applications Program asked the Eastman Kodak
Company to develop a CIR version of the digital camera.
Features considered important for remote sensing
applications include: (1) the camera system should consist
of a single camera body and lens assembly, (2) the camera
needs to be rugged, easy to use and relatively
inexpensive, and (3) the camera should include a GPS
interface to reference imagery with position data.
System Description
During 1994, Kodak developed a CIR digital camera system
for the Forest Service by modifying the camera firmware,
sensor, and software driver of a natural color DCS digital
camera. A DCS 200 camera was modified to create the first
CIR digital camera system. Subsequent versions of the CIR
digital camera will use the newer DCS 420 digital camera.
Both the DCS 200 and 420 digital cameras use a high
resolution 1,012(h) by 1,524(v) charge coupled device (CCD)
for imaging. The primary difference between the DCS 200
and 420 camera systems is dynamic range and image storage.
The DCS 420 has 12 bit dynamic range while the DCS 200 has
8 bit dynamic range for each spectral band. The DCS 200
uses an 80 MB internal hard drive that stores 50 image
files. The DCS 420 uses removable PCMCIA hard drives that
are available with different data storage capacities. A
200 MB PCMCIA hard drive can store 133 image files. The
DCS 420 can also obtain images in a rapid, continuous
sequence, which is an important feature for stereo
overlap.
The CIR digital camera
system obtains image data in the green, red and near
infrared spectral range by using a modified CCD and a band
pass filter attached to the camera lens. A band pass
filter ranging from 500 - 800 nanometers is used to limit
the spectral response to green, red and near infrared
reflected light. Although the CCD can be used to collect
image data up to 1.0 microns in the near infrared, we
found that for many resource applications it is preferable
to limit the near infrared band to 800 nanometers.
Different band pass filters can be defined for specific
imaging requirements.
The CIR digital camera
system provides a 24 bit color composite image through a
PC or Macintosh computer with image processing software. A
TWAIN driver combined with image processing software such
as Adobe Photoshop or Aldus Photostyler accesses image
data and does the color conversion. Each spectral band is
assigned to a primary color and displayed on a computer
monitor. The color conversion process assigns the near
infrared spectral band to red, the red spectral band is
assigned to green and the green spectral band is assigned
to blue. This spectral assignment creates a composite
color display similar to CIR photography. GPS position is
stored with the image data for general position
referencing. A serial data port receives GPS data in a
standard NMEA GPS data format. The TWAIN driver stores GPS
data with the image file. GPS data is displayed in the
camera information block along with the date, and camera
exposure data.
Preliminary Test Results
The camera was evaluated by collecting imagery on the
ground and from airplanes. Figure 1 displays a natural
color digital image taken of a forested area in Utah
during late fall. The scene includes a combination of
white fir (Abies concolor) and quaking aspen (Populus
tremuloides) exhibiting their fall coloration. Figure 2
displays a CIR digital image while figure 3 shows a CIR
photograph taken with a 70 mm Pentax camera using Kodak
2443 film. CIR imagery is considered more useful than
natural color imagery for highlighting the spectral
differences between healthy and dying vegetation. The
timber stand of white fir includes several dead trees that
are less distinct in the natural color image. In the CIR
imagery the older dead trees appear gray, the recently
killed trees are yellow-green, and the healthy coniferous
trees are dark red. Healthy deciduous vegetation normally
appears bright red in CIR imagery. However, since the
quaking aspen trees are beginning to turn color they
appear white or light pink in the CIR imagery.
An aerial CIR digital image
of a riparian site in northern California is displayed in
figure 4. The right side of the image includes a recent
burn area. Although this image was taken under a dark,
overcast sky, the digital camera performed quite well and
provided an image useful for delineating burned areas. The
burned vegetation appears as blue-gray color while the
healthy deciduous vegetation appears bright red. The
digital CIR camera can support fire recovery projects that
need aerial imagery quickly to access fire damage and
identify areas requiring erosion control or other
mitigation measures.
CIR film provides greater
spatial resolution than the digital camera. However, the
spectral response of the CIR digital camera system closely
matches CIR film. Image analysts who have many years of
experience interpreting color and CIR aerial photography
reviewed the digital CIR imagery. The consensus was that
the digital camera provided imagery spectrally very
similar to CIR photography. They also made the observation
that the CIR digital camera provided imagery with better
interpretability under adverse lighting conditions.
CIR film normally requires
special handling and processing to obtain reliable and
consistent results. CIR film has very narrow exposure
latitude. If the camera exposure is not within 1/2 f-stop
of the optimum setting, normal film processing will
produce less than desirable results. The digital camera
however, provides more latitude in exposure settings.
Digital cameras have greater dynamic range and provide
more flexibility in dealing with poor lighting conditions
and incorrect exposure settings. An individual digital
camera file can be adjusted with image processing software
to correct color balance and brightness.
A major advantage of
digital camera systems is the ability to use the data
directly with image processing and GIS systems. Image data
can be accurately georeferenced by transferring geodetic
control from orthophotographs, maps or previously
controlled aerial photography. Image data can also be
rectified using softcopy photogrammetry systems.
Conclusions
Digital camera systems will play an important role in
providing resource managers with timely and accurate
remote sensing data. A color infrared digital camera
system is a unique remote sensing capability that can
support a variety of ecosystem management applications
such as monitoring forest health, riparian condition, and
fire recovery. In addition, a color infrared digital
camera system will complement the use of other remote
sensor systems for updating GIS data bases, and performing
sample surveys and accuracy assessments.
About the Authors:
Tom Bobbe is a program leader with the USDA
Forest Service Nationwide Forestry Applications Program
located in Salt Lake City, Utah. He can be reached at
801-975-3663. Joe Zigadlo is an engineer working in
business development and custom imaging applications with
the Eastman Kodak Company located in Rochester, New York.
Note: The indication of commercial firms or
products in this article is for the convenience of the
reader. No endorsement is implied by the USDA Forest
Service.
Back
|