SIMULATION MODEL 
A simulation model with digital elevation and digital 
radiant value was prepared using national elevation data and 
LANDSAT MSS data respectively in the test area of Japan North 
Alps. 
Triple stereo images were digitally generated by computer to 
formulate 0.5 B/H ratio between the fore and vertical and the 
vertical and aft respectively. The pixel size was set up 10 
meters on the ground. 
The test area was composed of 260x260 pixels, where high 
mountains are located. 
The following cases were tested for comparisons of height 
accuracy and computing time. 
Case 1 Conventional Method with two rays 
Case 1-1 B/H ratio=0.5 
Case 1-2 B/H ratio=1.0 
A combination of a small window 3x3 and a large window 9x9 
was applied 
Case 2 Triplet Method 
Case 2-1 Independent Check Method 
Case 2-2 Simultaneous Check Method 
An allowed value for checking height accuracy was taken 60 
meters. B/H ratio was 0.5 for fore and vertical sight, and 
vertical and aft sight respectively. 
RESULTS OF SIMULATION 
Height accuracies for the four cases were shown in Table 1. 
Table 1 Height Accuracies 
Case 1 Conventional Method 
Case 1-1 B/H=0.5 6.64 m 117 m 
Case 1-2 B/H=1.0 24.79 m 493 m 
Case 2 Triplet Method 
Case 2-1 Independent 4.81 m 88 m 
Case 2-2 Simultaneous 5.45 m 38 m 
Computing times were shown in Table 2. 
227