1.. _raster.mg4lidar:
2
3================================================================
4MG4Lidar -- MrSID/MG4 LiDAR Compression / Point Cloud View files
5================================================================
6
7.. shortname:: MG4Lidar
8
9.. build_dependencies:: LIDAR SDK
10
11This driver provides a way to view MrSID/MG4 compressed LiDAR file as a
12raster DEM. The specifics of the conversion depend on the desired
13cellsize, filter criteria, aggregation methods and possibly several
14other parameters. For this reason, **the best way to read a MrSID/MG4
15compressed LiDAR file is by referencing it in a View (.view) file, which
16also parametrizes its raster-conversion. The driver will read an MG4
17file directly, however it uses default rasterization parameters that may
18not produce a desirable output.** The contents of the View file are
19described in the specification :ref:`MrSID/MG4 LiDAR View
20Documents <mg4lidar_view_point_cloud>`.
21
22MrSID/MG4 is a wavelet-based point-cloud compression technology. You may
23think of it like a LAS file, only smaller and with a built in spatial
24index. It is developed and distributed by Extensis. This driver supports
25reading of MG4 LiDAR files using Extensis' decoding software development
26kit (DSDK). **This DSDK is freely distributed; but, it is not open
27source software. You should contact Extensis to obtain it (see link at
28end of this page).**
29
30Example View files (from View Document specification)
31-----------------------------------------------------
32
33Simplest possible .view file
34~~~~~~~~~~~~~~~~~~~~~~~~~~~~
35
36The simplest way to view an MG4 file is to wrap it in a View (.view)
37file like this. Here, the relative reference to the MG4 file means that
38the file must exist in the same directory as the .view file. Since we're
39not mapping any bands explicitly, we get the default, which is elevation
40only. By default, we aggregate based on mean. That is, if two (or more)
41points land on a single cell, we will expose the average of the two.
42There's no filtering here so we'll get all the points regardless of
43classification code or return number. Since the native datatype of
44elevation is "Float64", that is the datatype of the band we will expose.
45
46::
47
48   <PointCloudView>
49      <InputFile>Tetons.sid</InputFile>
50   </PointCloudView>
51
52Crop the data
53~~~~~~~~~~~~~
54
55This is similar to the example above but we are using the optional
56ClipBox tag to select a 300 meter North-South swatch through the cloud.
57If we wanted to crop in the East-West directions, we could have
58specified that explicitly instead of using NOFITLER for those.
59Similarly, we could also have cropped in the Z direction as well.
60
61::
62
63   <PointCloudView>
64      <InputFile>Tetons.sid</InputFile>
65      <ClipBox>505500 505800 NOFILTER NOFILTER</ClipBox>
66   </PointCloudView>
67
68Expose as a bare earth (Max) DEM
69~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
70
71Here, we expose a single band (elevation) but we want only those points
72that have been classified as "Ground". The ClassificationFilter
73specifies a value of 2 - the ASPRS Point Class code that stipulates
74"Ground" points. Additionally, instead of the default "Mean" aggregation
75method, we specify "Max". This means that if two (or more) points land
76on a single cell, we expose the larger of the two elevation values.
77
78::
79
80   <PointCloudView>
81      <InputFile>E:\ESRIDevSummit2010\Tetons.sid</InputFile>
82      <Band> <!-- Max Bare Earth-->
83         <Channel>Z</Channel>
84         <AggregationMethod>Max</AggregationMethod>
85         <ClassificationFilter>2</ClassificationFilter>
86      </Band>
87   </PointCloudView>
88
89Intensity image
90~~~~~~~~~~~~~~~
91
92Here we expose an intensity image from the point cloud.
93
94::
95
96   <PointCloudView>
97      <InputFile>Tetons.sid</InputFile>
98      <Band>
99         <!-- All intensities -->
100         <Channel>Intensity</Channel>
101      </Band>
102   </PointCloudView>
103
104RGB image
105~~~~~~~~~
106
107Some point cloud images include RGB data. If that's the case, you can
108use a .view file like this to expose that data.
109
110::
111
112   <PointCloudView>
113      <InputFile>Grass Lake Small.xyzRGB.sid</InputFile>
114      <Band>
115         <Channel>Red</Channel>
116      </Band>
117      <Band>
118         <Channel>Green</Channel>
119      </Band>
120      <Band>
121         <Channel>Blue</Channel>
122      </Band>
123   </PointCloudView>
124
125Writing not supported
126---------------------
127
128This driver does not support writing MG4 files.
129
130Limitations of current implementation
131-------------------------------------
132
133Only one *<InputFile>* tag is supported. It must reference an MG4 file.
134
135The only *<InterpolationMethod>* that is supported is *<None>*
136(default). Use this to specify a NODATA value if the default (maximum
137value of the datatype) is not what you want. See View Specification for
138details.
139
140There is insufficient error checking for format errors and invalid
141parameters. Many invalid entries will likely fail silently.
142
143See Also:
144---------
145
146-  Implemented as *gdal/frmts/mrsid_lidar/gdal_MG4Lidar.cpp*
147-  :ref:`MrSID/MG4 LiDAR View Document
148   Specification <mg4lidar_view_point_cloud>`
149-  `Extensis web site <http://www.extensis.com/support/developers>`__
150
151.. toctree::
152   :maxdepth: 1
153   :hidden:
154
155   mg4lidar_view_point_cloud
156