The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part BI. Beijing 2008 
For above information, statistics them before processing and 
save this index data to one XML file. Using this index data, we 
can seamlessly management and view a large number of point 
clouds. In past, it’s difficulty to realize data distribution, 
because the size of data is too large and beyond the computer. 
But using geocoding index, we can draw the smallest external 
rectangular for every file to realize data distribution. And this 
method is simplicity and intuitive to global view as below Fig 
shown: 
□ 
■trip 1 
[□ 
□ 
□ 
□ 
□ 
□ 
□ B 
strip 2 
□ 
□ 
□ 
□ 
strip n 
□ 
□ 
□ 
□ 
kJ 
stlp 
In other hand, in point clouds, scan angle is ignored usually. 
Scan angle equivalent to scan line for data. And in practice, 
many data organizations using scan line as the smallest unit. 
Because the file position of data in one scan line is known, it is 
not necessary to load data by files. Through select a geodetic 
area to load dataset which from many files and part of some 
files, the loading formula is: 
In rapid viewing large number of data, simplifying the data can 
be separated by a number of scanning lines directly reading data. 
This seamless point cloud data management methods 
significantly improve the efficiency of the man-machine 
cooperation. 
3.3 Point cloud pyramid 
The information from human eyes, are obtained based on multi 
resolution, which also accords with the human cognitive things 
by coarse-to-fine features, and has used in image processing. 
LiDAR data is also on a description of the objective world, we 
can still use the point cloud pyramid organizational mechanisms, 
and its algorithm is as follows: 
Every point cloud file relates to a point cloud index files. This 
index file contains n levels, as below (4 layers): 
Every level of index files stored in the information below, 
taking 3*3 as example: 
Recording the coordinate scope for every block; 
Recording the offset position in file for this block data. 
In LOD algorithm, far from eye point, using 3*3 index; close 
from eye point, using 81*81 index. 
4. ADAPTIVE VISUALIZATION FOR AIRBORNE 
LIDAR 
In resolving session of airborne LiDAR, data visualization is 
usually used in bore-sight calibration, LiDAR’s data 
classification, generating DSM (Digital Surface Model), object 
reconstruction by fusing LiDAR Data with Photogrammetry, 
and so on. But in these sessions, not all operation need display 
full data. For instance, in bore-sight calibration, we only need 
profile of LiDAR’s data to feedback the setting of computing 
parameters. So a new adaptive visualization algorithm is 
proposed to improve customer experience and quicken data 
resolving. 
In bore-sight calibration: 
4.1 Pitch 
Choose to fly through the cuspate roof back and forth to 
analyze the airline data. Draw a block that contains the area of 
the root top among the generated point cloud data, and build up 
two surfaces. Draw a profile plumbing the roof top in both of 
the two surfaces to display the profile of the two surfaces 
together. Measure the errors in the horizontal orientation of the 
two profiles, and then divide the flying height to come up with 
a validation value. See the theory in the diagram. 
Re-input the value into the LiDAR pre-handling software to re 
generate point cloud data. Use the same way to re-build profile. 
If those two profiles match well, the validation value for 
Pitching is proved to be valid. If not, go back and redo these 
steps to revise based on the previous validation value until the 
two profiles match completely. According to the steps above, 
not all the profiles data are used in the revising process. 
real ground 
recording point