1. A new laser scanning reference text by Shan and Toth appears to be a must read.
  2. Intensity is often misused in the laser scanning industry.
  3. There are a number of factors that affect the intensity of the return signal.

I have just received what appears to be an excellent laser scanning technical resource, Topographic Laser Ranging and Scanning, by Jie Shan and Charles Toth. I mentioned this a few weeks ago. More on this once I have a chance to review.

One term that I could not find discussed in much detail was intensity, so I checked with my ASTM E57.04 colleague Kevin Ackley from Course Six, Inc. he provided me with the following insights:

The term intensity, I think, is frequently misused. In radiometry (measurement of light) it means irradiance, which is optical power per unit area (watts per square meter). The imprecise definition, which most people are using is : how much light (watts) did the instrument receive.

There are a number of factors that affect ” intensity”:

  1.  How much light was sent out?
  2. Range to the target
  3. Attenuation of beam by atmosphere
  4. Reflectance of target (at the appropriate wavelength)
  5. Incident angle of light on the target surface
  6. Size of receiver aperture
  7. Efficiency of receiver optics at focusing the return light onto receiver sensor
  8. Specularity of the target surface

An instrument manufacturer would want to compensate for some of these effects so that the  “intensity” number is more meaningful. Items 1, 6, and 7 are the easiest – they are internal to the instrument. Item 2 should be known after the range calculation is complete. But items 3, 4, 5, and 8 are dependant on conditions remote from the instrument. My guess is that most instruments would compensate for 1, 2, 6, and 7. The remaining factors are what the “intensity” is really characterizing.

Side note: Surprisingly (to me), the angle of incidence of the transmitted beam onto the target surface does effect the return signal strength (item 5), This assumes a monostatic configuration (co-located transmit and receive). The return power is proportional to cosine of incident angle. So as the surface normal of the target tilts away from the direction of the incoming light the return power is reduced.

Thanks Kevin. That helps to understand why intensity is not all that valuable an attribute.

This entry was posted in Education, Research, Technology, training and tagged , , . Bookmark the permalink.

7 Responses to Intensity

  1. While we have found a lot of variability in the quality of airborne LiDAR derived intensity, we find it incredibly valuable for automated land cover mapping. Impervious surfaces are much easier to extract from LiDAR intensity data than comparable resolution color infrared imagery. We believe this is because the contrast is typically more pronounced in LiDAR intensity data than in the imagery. By using contrast to surrounding features one gets around the issues associated with the values being consistent across an intensity dataset (the advantage of object-based image analysis). That being said, we really don’t have a good understanding of LiDAR intensity data from a user’s perspective. I would be interested to know if Shan and Toth cover this in their book.

  2. I added a post to our blog showing how we use intensity data. Been meaning to do this for some time, thanks for the (virtual) push.

  3. Raj says:

    VLS’s 2 tools Feature Analyst and Lidar Analyst, have been combined to create workflows which allow you to extract features from the intensity data in LIDAR.
    One example: LIDAR Analyst is used to extract bare-earth.
    Feature Analyst is then used to extract roads from the intensity, using the bare-earth to mask out non-ground areas.
    I knew that there used to be a paper that I had written on it a while back that I cant find online anymore. But check out this workflow document –

    In addition, RoadTracker – a plugin for Feature Analyst from GeoEye, allows you to perform semi automated road extraction from lidar intensity images.

  4. Great information! I recall having a similar discussion with a friend not too long ago on intensity and how exactly does the receiver relate return signal strength through voltage to actual intensity values we observe in the data. It seems that the actual value we get from the system is a relative measure of the integration of received power over time for the given return signal (i’ll ask around and see if I can get a straight answer). As the article explains, this value is a combination of several factors some of which, incidence angle, surface specularity, atmospheric factors, are too difficult and simply not feasible to compensate for all their effects. This is a major difference between radiometric ally calibrated data as we see in some passive systems and spectral libraries. As such, lidar intensity as we know it is a relative measure meaning that given a surface with the same reflective properties, the intensity value you observe will vary temporally because of the inability to compensate for the factors mentioned above. In addition, system “look up tables” used for processing to compensate for range walk due to differences in very weak and strong returns will affect the output intensity value. Thus, intensity will vary within and across surveys. Although it is a relative measure, it has no doubt found great utility as a feature as other posters have pointed out. We’ve had great success in using lidar intensities for delineation of the high-water line along beaches, building and tree segmentation, as well as other uses in classification. And it appears that there is an ever increasing interest in intensity and the push towards methods to compensate more for incidence angle, atmospheric effects, etc.. There was an excellent and fairly recent paper that discussed this issue and attempts for better calibration models for intensity. I’ll see if I can find it.

    For our work, we always normalize the intensities based on the square of the range. This provides some control on the range attenuation across a given survey. We developed code to do this though nowadays it seems most vendors include this capability in their processing software, such as Optech’s DashMap. Anyhow, if interested below is a link to a technical report on range normalization including original source code:

    Additionally, the link below has several reports on lidar basics and processing as well as a few links/references to some of our papers on using lidar as a feature.

    Thanks for the great site and wealth of lidar information. Lidar rocks so please keep it coming. I just came across this site and enjoy it very much.

Leave a Reply

Your email address will not be published. Required fields are marked *


This site uses Akismet to reduce spam. Learn how your comment data is processed.