intensity differences between flightlines

[Martin Isenburg]

Hello,

one thing that has been on my RADAR aeeehhh LiDAR for a while is to have a closer look at the different intensity values that occur in LAS/LAZ files. Several times already I have noticed (and maybe you have too) very obvious differences in intensity for the same area in overlapping flightlines. The first time I noticed that was when I visited Ataneo de Naga university in the Philippines and it was documented in these pictures. This really prevent us from using the intensities to compute meaningful forestry metrics with lascanopy or FUSION:

http://www.facebook.com/LAStools/photos/846786945383751
http://www.facebook.com/LAStools/photos/846786915383754
http://www.facebook.com/LAStools/photos/846786895383756

I computed *single return* (!!!) intensity histograms for the two strips that have such different brightness and the result is rather odd. The brighter strip has a wide and flat distribution with spikes at very particular values whereas the darker strip has a nice peak like I would expect it.

Can anyone explain these distributions? I am quite sure that they are both from an Optech Gemini sensor.

Is anyone willing to share similar data with me where overlapping flightlines have noticable difference in the brightness values of the intensities?

Regards,

Martin

PS: Here how to compute such histograms:

D:\LAStools\bin>lasinfo -i martin_intensity_issue_3_5.laz ^

                                      -keep_first -keep_point_source 3 ^

                                      -nh -nv -nmm ^

                                      -histo intensity 4

lasinfo (160124) report for martin_intensity_issue_3_5.laz

intensity histogram with bin size 4

  bin [0,4) has 34473

  bin [4,8) has 39848

  bin [8,12) has 37535

  bin [12,16) has 39918

  bin [16,20) has 41549

  bin [20,24) has 44217

  bin [24,28) has 50714

  bin [28,32) has 57529

  bin [32,36) has 63845

  bin [36,40) has 69448

  bin [40,44) has 76207

  bin [44,48) has 80249

  bin [48,52) has 84754

  bin [52,56) has 85318

  bin [56,60) has 87081

  bin [60,64) has 84428

  bin [64,68) has 79698

  bin [68,72) has 71528

  bin [72,76) has 60378

  bin [76,80) has 47530

  bin [80,84) has 35339

  bin [84,88) has 24389

  bin [88,92) has 16406

  bin [92,96) has 9732

  bin [96,100) has 5994

  bin [100,104) has 3059

  bin [104,108) has 1507

  bin [108,112) has 674

  bin [112,116) has 310

  bin [116,120) has 128

  bin [120,124) has 57

  bin [124,128) has 23

  bin [128,132) has 12

  bin [132,136) has 9

  bin [136,140) has 4

  bin [140,144) has 1

  bin [148,152) has 1

  bin [160,164) has 3

  bin [164,168) has 1

  bin [172,176) has 1

  bin [180,184) has 1

  bin [220,224) has 1

  bin [280,284) has 1

  average intensity 47.3731 for 1333900 element(s)

D:\LAStools\bin>lasinfo -i martin_intensity_issue_3_5.laz ^

                                      -keep_first -keep_point_source 5 ^

                                      -nh -nv -nmm ^

                                      -histo intensity 4

lasinfo (160124) report for martin_intensity_issue_3_5.laz

intensity histogram with bin size 4

  bin [0,4) has 9271

  bin [4,8) has 8895

  bin [8,12) has 7055

  bin [12,16) has 6903

  bin [16,20) has 6430

  bin [20,24) has 5651

  bin [24,28) has 5070

  bin [28,32) has 4815

  bin [32,36) has 4682

  bin [36,40) has 4601

  bin [40,44) has 4559

  bin [44,48) has 4711

  bin [48,52) has 4694

  bin [52,56) has 4916

  bin [56,60) has 4939

  bin [60,64) has 5181

  bin [64,68) has 5102

  bin [68,72) has 5518

  bin [72,76) has 5608

  bin [76,80) has 6027

  bin [80,84) has 6213

  bin [84,88) has 6415

  bin [88,92) has 6721

  bin [92,96) has 6891

  bin [96,100) has 7765

  bin [100,104) has 7422

  bin [104,108) has 7721

  bin [108,112) has 8018

  bin [112,116) has 8247

  bin [116,120) has 8249

  bin [120,124) has 8831

  bin [124,128) has 8740

  bin [128,132) has 8907

  bin [132,136) has 9125

  bin [136,140) has 9222

  bin [140,144) has 15024

  bin [144,148) has 9663

  bin [148,152) has 9418

  bin [152,156) has 9659

  bin [156,160) has 9728

  bin [160,164) has 48460

  bin [164,168) has 9886

  bin [168,172) has 9883

  bin [172,176) has 9983

  bin [176,180) has 9994

  bin [180,184) has 53363

  bin [184,188) has 7464

  bin [188,192) has 1080

  bin [192,196) has 17

  bin [196,200) has 1

  bin [200,204) has 32511

  bin [220,224) has 19935

  bin [240,244) has 9644

  bin [260,264) has 4139

  bin [280,284) has 1530

  bin [300,304) has 487

  bin [320,324) has 103

  bin [340,344) has 23

  bin [360,364) has 4

  bin [400,404) has 2

  bin [420,424) has 5

  bin [440,444) has 2

  bin [480,484) has 1

  bin [520,524) has 3

  bin [540,544) has 2

  bin [580,584) has 3

  bin [620,624) has 3

  bin [640,644) has 1

  bin [720,724) has 1

  bin [740,744) has 1

  bin [780,784) has 1

  bin [800,804) has 1

  bin [840,844) has 1

  bin [860,864) has 1

  bin [920,924) has 1

  bin [980,984) has 1

  bin [1040,1044) has 1

  bin [1100,1104) has 1

  bin [1120,1124) has 2

  bin [1260,1264) has 1

  bin [2080,2084) has 1

  bin [2140,2144) has 1

  average intensity 130.732 for 505151 element(s)

[Edgar Veldman]

Hi Martin,

Since intensity is the energy value of the returned pulse, but rescaled to an 8 or 16 bit value, this value can be correlated with the flying height; the longer the pulse has to travel, the more of its energy is lost. 

In your screenshots I can clearly recognize the same objects in both flight-lines, but at a different ‘scale’, and as you can see the narrower swath (captured from a low run) is brighter, and the wider swath (captured from a higher run) is much darker.

There may be many more factors that contribute to an intensity difference, such as weather, bad calibration of the sensor, wetness of the targets, the difference between outgoing angles to the same targets, but in this extreme case I think the flying height is the main contributor.

Edgar

 

[Martin Isenburg]

Hello Edgar,

You cannot see the entire swath in this (tiled) example but I believe the intensity difference is indeed due to different flying heights. And I would like to be able to make the two intesities be more conforming. However, what I absolutely do not understand is the histogram. I would expect the first return intensity histogram to look similar with the peak elsewhere and the distribution maybe widened or squished a bit. But what we get for the higher intensity distribution is something really funky. A few values have insane spikens and then the values suddenly show a sharp second peak above 200.

[...]

  bin [156,160) has 9728
  bin [160,164) has 48460
  bin [164,168) has 9886

[...]

  bin [176,180) has 9994
  bin [180,184) has 53363
  bin [184,188) has 7464

[...]

  bin [192,196) has 17
  bin [196,200) has 1
  bin [200,204) has 32511
  bin [220,224) has 19935

[...]

These artifacts in the distribution are clearly visible in the histogram. I cannot explain them with a higher flying height ... anyone seen that from their Optech Gemini / LMS Output?

Martin

--
Download LAStools at
http://lastools.org
http://rapidlasso.com
Be social with LAStools at
http://facebook.com/LAStools
http://twitter.com/LAStools
http://linkedin.com/groups/LAStools-4408378
Manage your settings at
http://groups.google.com/group/lastools/subscribe

[Edgar Veldman]

Martin,

For this example, it almost looks like the white patch (a cloud?) has affected the rest of the intensities. It could be that the Optech post-processing software normalises the intensity values based on the min/max observed in the entire swath, and therefore smudges the majority of the otherwise valid data down to the lower intensity values? I have seen post processing software do many undesired operations with lidar or imagery data, this could be one of those operations ...

Would clamping and rescaling the intensity values bring out the detail again in the lower intensities, or is there no detail left at all?

Edgar

[Martin Isenburg]

Hello Edgar,

good to have you back. (-:

Yes, all the detail in the darker flightline is still there. But that is the one that has the *nice* histogram of intensities. Here are the instructions how to rescale the intensities to bring out the details for visualization such that you get the enhanced image that indeed has a cloud:

http://www.facebook.com/LAStools/photos/846786865383759

However, this is *not* the histogram of intensities I was talking about. I am talking about the *ugly* histogram of intensities that comes from the brighter flightline.

Regards,

Martin

[Edgar Veldman]

Hi Martin,

I was focusing on the dark swath, not the bright one, my bad.

Without having a copy of the data its difficult to determine where these spikes come from, do you have an idea about the objects or terrain type these spikes originated from? Also, when running the histogram on all data (not single returns only) do you see the same gap before the second peak, or is there a smoother transition between the values?

thanks,

Edgar

[Evon Silvia]

Unfortunate that the group that acquired the data didn't refly the line because of the cloud. That should be SOP for LiDAR acquisition.

Optech does have some blackbox intensity normalization options available in the more recent versions of their postprocessing software. I believe that they have the option to normalize for range, which is better than flying height for low AGL, but I don't know what else there are doing. It seems plausible that they would also rescale the entire line to utilize the full 12bit range of their sensors, or possibly even the full 16bit range.

Been a while since I've played with it since I run my own normalization routines, but that's what it looks like to me.

Evon