The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008 
between high resolution panchromatic images and high- 
spectral images(LI Jun, ZHOU Yue-Qin and LI 
DeRen,1999), spot images and TM images(WANG Zhijun, LI 
Deren and LI Qingquan,2001), IKONOS images and multi- 
spectral images(WANG Zhijun, LI Deren and LI 
Qingquan,2001), QuickBird images(L IU Chun and CHEN 
Neng, 2004), Sea Ice Remote Sensing Image(WU Kui-qiao, 
WANG Hu, HUANG Run-heng and LIU Jian-qiang,2005). 
Besides these technologies, Support Vector Machines(ZHAO 
Shuhe , FENG Xuezhi , DU Jinkang and LIN Guangfa, 
2003)and Genetic Algorithm(TONG Xiao chong, ZHANG 
Yong sheng and BEN Jin, 2006) are also used for image fusion. 
As discussed above, image fusion had been studying for many 
years. However, rarely research results can be seen focus on 
fusion between LIDAR points and images. 
2. DATA SET AND DATA PROCESS 
2.1 Data 
The data used in this research represent a region in Yantai, 
Shandong province of east China and all the data are supplied 
by North China Sea Branch of SOA. And the data are 
generated by ALS 40 LIDAR system. There are over 8 flight 
strips in the original data set. The data acquisition time is about 
2.5 hours. Since the huge data volume, two flight strips are 
selected for research and experiments. The rough latitude is 
about 37° 35' 53.74" and the longitude is about 121 ° 23' 
06.18" . 
After post process of LIDAR, the total number of points in 
these two flight strips is about 6,570,000, however the number 
of Ortho-photomaps is 22. Obviously, the data is still too huge 
for research. In this paper, the sub-area with the coordinate 
scale (621290, 416475)-(622666, 4163298) are selected for 
further data process and fusion research. Thus, the data in this 
area, including the data points and the Ortho-photomap, are 
split from the processed data. The area of this region is about 2 
km 2 ■ We can also find that there are hills, lakes, sea, buildings, 
roads in this region. The resolution of this image is 2052*2178 
and the pixel resolution is 0.667 m . The number of the points 
in this region is over 290,000. Fig.l shows the location and the 
Ortho-photomaps of this region. 
Fig.l Location and Ortho aerial photo of research area 
2.2 Data Component of Points Cloud 
The data set from field survey is processed by some software 
which is supplied by LIDAR systems. With some steps, like 
post GPS process, interpolation of IMU data, integration of 
GPS data and IMU data, integration with laser data, instrument 
and temperature correction, intensity correction, data 
projection, the data points, which is called Points Cloud, will 
be gathered in a file. Usually, the points cloud is stored in a las- 
file. With the help of TerraScan and TerraMoulder, the las-file 
can be changed into a text file. Thus, the content of points 
cloud and the main data component of a text file is listed as 
below: 
Coordinate _ X, Coordinate _ Y, Coordinate _Z, Intensity 
Here, Coordinate _ X,Coordinate _Y,Coordinate _Z indicates 
the geometry of a certain point, while intensity indicates the 
optical information of each echo. 
The coordinate system used here is the Guass-projection with 
WGS84 global reference. During the data process, the same 
coordinate system is adopted in both points cloud and Ortho 
photomaps. So the reference of these two data set are nearly 
the same. Since there’s no public ground control points(GCPs) 
between images and points, the matching between two data set 
is not conducted. If there’s enough GCPs, the matching is 
needed for experiment. 
The following Figure 2 shows the points used in this paper. 
Fig.2 Points Cloud(298,699 points) 
2.3 Interpolation of Points Cloud 
Points cloud gives the coordinates and optical information in 
vector format, while the Ortho-photomaps gives the spectral 
and geometric information in raster format. Traditionally, 
there’s no direct method for data integration between vector 
data and raster data. In order to integrate the information of 
points cloud and that of raster images, two methods can be 
used to achieve the purpose. The first one is to interpolate the 
points cloud in vector format into raster format. Then some 
useful fusion approaches, such as IHS transform and PCA, are 
selected to integrate the two data. The second method is to 
integrate every point in points cloud with raster pixel according 
to the coordinate relationship. Obviously, the result by first 
method is a raster file and it covers the whole area. The result 
by second method is a vector file and the volume of result data 
is the same with original points cloud. But it may not cover the 
whole area. In this paper, the two methods are both used to 
implement the fusion process. 
310