The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B5. Beijing 2008 
612 
orientation elements of exterior and interior, so the collinearity 
equation of photogrammetry can not be used to calculate the 
these parameters. In here, we use the DLT to compute them. 
3.1 DLT Principle 
DLT basic equation is come from collinearity equation of 
photogrammetry as shown in the equation (1). From this 
equation we can conclude that DLT displays the linearity 
relation between image coordinates and object coordinates in 
ground coordinate system. The 11 parameters of this equation 
can be directly computed by using the selected corresponding 
points. But, for improving the precision of the registration, there 
must be redundant observation to take part in least squares 
adjustment. The coefficient elements of exterior and interior are 
computed according relation equation between them and 11 
parameters. Then, we can also use the collinearity equation to 
register digital image and LIDAR data. 
l x X + l 2 Y + I 3 Z + / 4 A 
l 9 X + l w Y + l\\Z + 1 (1) 
LX + LY + LZ + L 
y + _5 1 z — = 0 
l 9 X + I l0 Y + I U Z +1 
where x, y = image coordinates 
X, Y, Z= object coordinates in ground 
coordinate system 
4. DATA FUSION AND TEXTURE MAPPING 
This section introduces data registration, texture mapping, 
displaying of textured point cloud and TIN model and the 
approach of making the orthophoto map. 
4.1 Data Registration and Fusion 
According to above calculation, we have got orientation and 
lens distortion parameters. Thus, the rigorous collinearity 
equation can be obtained. Then, texture coordinates of each 
point of point cloud and TIN model can be computed by using 
this equation. But, the calculated texture coordinates values 
commonly are not integers. Some interpolations should be 
applied. The paper uses the best neighbour point method. 
Calculating the distance of the calculated point to the four point 
border upon it, and then, choosing the point of the shortest 
distance away from the four points as the last value of texture 
coordinates. After that, we actualize the registration and fusion 
between digital image and LIDAR data. Thus, each LIDAR 
data point has its 3D and image coordinates. 
4.2 Texture Mapping 
Paper uses the texture mapping technology of OpenGL to 
realize the display of textured point cloud and TIN model. 
Because the each point of point cloud has got its 3D and texture 
coordinates, using the point displaying function of OpenGL to 
show the point and its texture. Figure 2 is the result. The left is 
the digital image, and the right is the textured point cloud. 
3.2 Using DLT of Iterative Least Squares to Calculate the 
Precise Values of Parameters 
Under the circumstance of redundant observation, paper utilizes 
the iterative least squares to computer the parameters. Firstly, 
by using the function equation of DLT and 6 pairs of 
corresponding points, we can get 11 linearity equations. From 
these equations, we can computer the initial values of 11 
coefficients. There should be noticed that these corresponding 
points had better to be distributed around the image and model 
in order to obtain a better initial values. For the purpose of 
camera calibration, there add an optics lens distortion 
coefficient. In fact, there are lots of lens distortion coefficients, 
but, this added has satisfied the request. Next, more than 6 pairs 
of corresponding points can get the coefficient and constant 
matrix of error equations. According to the normal equation, 11 
coefficients and lens distortion coefficients can be obtained. 
This is an iterative process, so we consider the photo focus 
difference of adjacent twice calculations as the iterative basis, 
and stop the calculation when the difference is lesser than 0.01 
mm. There should be known that the photo focus is calculated 
by using the 11 coefficients. Finally, the orientation elements of 
exterior and interior can be calculated by using the conversion 
relation equations about 11 coefficient and these elements. 
In this paper, terrestrial LIDAR HDS4500 is applied to acquire 
point clouds, and high resolution digital camera is applied to 
obtain texture data. The system used 19 pairs of corresponding 
points to take a test. According to the DLT of iterative least 
squares, the orientation elements of exterior and interior and 
lens distortion parameter have been calculated, and prepare for 
the data registration. 
Figure 2 Textured Point Cloud 
Figure 3 Textured TIN Model