The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008 
470 
Coordinate systems 
Through the above model, we get the ground 3D point clouds 
following formula: 
X 
X" 
y 
= 
Y„ 
^WGS-m p,o,k)^Rcf (IMif) 
Y l 
z 
tVGS-84 
A. 
IVGS-84 
A. 
Ref is the reference coordinate system for IMU, its relationship 
between IMU coordinate system is restrictionist for resettlement 
of IMU. So if we set up the conversion relations in POS control 
system, the output will be in reference coordinate system for 
IMU. 
[X! Y l Z l ] is coordinates of point clouds in laser system, and 
its value is calibrated. 
2.2 The mechanism of airborne LiDAR's data process 
The process of airborne LiDAR is shown in Fig below: 
1. Decoding, extract the raw scanning data according to 
data types, but in data file of any type ,the time sequence 
is same as raw data. Such as remote station GPS, INS, 
laser scanning data(range, angle, intensity). 
2. Using GPS processing software to resolve the master 
station and remote station GPS data. 
3. Using POS processing software to combine GPS and 
IMU with smoothing algorithm. 
4. Matching multiple data to compute coordinates, in 
this stage, the main operation is interpolation by time. 
5. Computing coordinates of points, it contains 
correcting, coordinates rotation and computing XYZ 
values. 
6. Outputting data, generating the recognized format file 
for point clouds. 
2.3 Data types of Airborne LiDAR 
In accordance with its data Solution to classify the role of all 
kinds of data. The data type of airborne LiDAR contains: 
a) Raw Scanning Data 
Raw scanning data; 
Navigation sbet file from POS processing software; 
Survey area data; 
b) Middle Processing Data 
Laser scanning data (range, angle, intensity); 
Time synchronization data; 
Equipment status and environment parameters data; 
Log data in flight; 
c) Outputting data 
ASCII format XYZI file(X, Y, Z, Intensity); 
The internationally popular laser point cloud data files 
LAS, developed by the ASPRS. 
3. MANAGE POINT CLOUDS 
One LiDAR (laser radar) sensor generates over a hundred 
millions of XYZ coordinate parameters. These great amount of 
sample points are described “ non-organized ” . When those 
LiDAR points are re-built, they become “Cloud” . How to 
organize these non-organised points? 
3.1 Problems 
Those hundreds million points course calculation ball-up, 
because it is impossible to store all the data into the memories 
of the most complicated computer. Therefore, the data should 
be the hard disks which are in larger volumes but slower speed. 
Handling the large number of data within the hard disks and 
RAM instead of within computers has some disadvantages in 
the manipulation. Thus, most practical algorithms use section 
algorithm to rupture the cloud point into a series of non- 
overlaid second level cloud which includes a few sample points. 
In every section, the data interpolation is independent. Based on 
this principal, many based on section methods are developed, 
including the simple rupture and ruptures based on Voronoi 
line-drawing figuring method. One algorithm to match the data 
exchange can minimize the time period of visiting hard disk 
which also compress the time of running. People are still 
searching for algorithms that can deduce the time of data 
exchange. 
3.2 Geocoding Index 
Common characteristics of all point clouds are geodetic 
coordinates. So we should not organize data through files, using 
geodetic location of every point to manage all data. In this case, 
for every file below information is must known: 
a) In the XY plane the smallest external rectangular for 
the data in one file; 
b) In LiDAR system every scan line’s tail data offsets to 
file starting point; 
c) The storage path of every file, this path also can in 
LAN.