SO503 Spring 2009

SO503 Spring 2013

Matlab and LIDAR Point Clouds/3D graphics

Starting in MATLAB 7 (R14), the syntax is:

%{
   Stuff to be commented out
%}

You can also highlight a section of code and type control-r to comment out the code -- this will place a percent symbol (%) at the beginning of the line. Typing control-t will uncomment the lines by removing any percent symbol that is the first non-blank character on the line.

If you are using an even older version, use this:

if 0
  commented out code 
end

LIDAR point clouds come in a wide variety of formats. The simplest are ASCII text files, with three values per line, x, y, and z. It is quite likely, however, that that files will be too big to deal with in Excel, at least the older versions of the program. There are two variants:

Whether or not there is a header line, with the names of the columns. You can add or remove this with an editor like Notepad or Wordpad, although Matlab does not appear to be very fussy about this.
What separates the values (the delimiter, which can commonly be a comma, a space, or a tab). You should look up the dlmread command in the Matlab help to see how you can set the correct delimiter. You can also do a global search and replace in an ASCII editor, but this is likely to be painfully slow.

For serious use, the most common format is LAS, which is binary and substantially more compact that these ASCII files. LAS files can have hundreds of millions of points (up to 15 points per square meter), and gigabytes in size.

You will use the following Point Clouds:

v1-thin5.txt: from a historical ship. (178,319 points; this was thinned by MICRODEM from a file 5 times larger)--try a grid size of 0.1
toronto.csv, from a Canadian city (65,535 points; this was truncated by Excel from a larger file)--try a grid size of 1
2008_NH_Portsmouth_30464_raw.txt (2326 points) --try a grid size of 1

We will be doing Lab 1 in the reading, which starts on page 32. You have the data file already, so you can jump to the middle of page 35 for directions on using Matlab. We do not have an m file to start with, but you can use the code on page 38 as a starting point. Note that these procedures will be happiest with data imported as a matrix, which is a little different than we have used to date when we have used vectors.

You will ultimately produce Figures for all three datasets; if your Matlab is slow, you might think about which data set will be best (fewest or most points?) to use while you are debugging your code.

Create the following figures (you might have to check the help on the parameters to use with each; there are samples of all 5 below):

surfl
surface
surfc
trimesh
scatter3

You can open the three files in MICRODEM (latest version only) with the Point Clouds menu choice. You can also look at a Baltimore LIDAR file in MICRODEM; its 2.8 million points will probably choke Matlab.

Turn in for this lab:

At least 5 figures, one from each of the commands above, with all three of the data sets represented among your graphs so that I can be sure you successfully imported each data set. You should insure that you export as dumb pictures, and not as an embedded object, or you will have a painful slowdown every time your Word screen tries to redraw. If that occurs when I try to grade your paper, it will come back to you for correction.
Your matlab source code, showing how you modified the code to deal with the three different input files.
A discussion about whether you think the three data files are essentially identical, or if they have geometric differences which determine which of the visualization options best depicts the data.
A comparison of the advantages and disadvantages of using Matlab to visualize this kind of data, compared to a more specialized program.

Links for this lab only available within USNA

Matlab figures for surfl, surface, surfc, trimesh, and scatter3 for the smallest data set

MICRODEM Images for the two larger data sets