displacements, that digitizing would include pointing error, 
and the displacement computed from the digitized coordinates 
would include the effects of pointing error. For purposes 
of testing the matching algorithms, features were chosen 
that contained discrete points and edges to reduce this 
source of error as much as possible. 
6. CONCLUSIONS 
As indicated by the feature matching results, the boundary 
detection and matching algorithm developed here can compute 
anatomic feature displacement to sub-pixel accuracy on 
Sequential cranial x-ray films. The use of a feature 
extraction algorithm serves to reduce the background X-ray 
noise, and produces a discrete data set for feature 
matching. The use of a six parameter boundary transformation 
successfully  compensates for the non-orthogonality of 
feature growth over the two year period. 
Although the  accuracies are not high by photogrammetric 
standards, they are satisfactory for biostereometric 
Purposes, and are partly a function of the low resolution of 
the array camera used. The techniques described in this 
report would find application in feature matching on aerial 
photographs where features have undergone distortion due to 
morphological change. However, it is limited to those cases 
where discrete feature boundaries can be extracted. 
7. REFERENCES 
/1/ Ackermann, F.: Digital Image Correlation : 
Performance and Potential 
Applications. 
The Photogrammetric Record, Vol 
1, No. 64, (1984), pp. 429-440. 
/2/- Curry, S.., Baumrind, Analysis of Stereo Cranial X-Rays 
Using Digital Images. 
Close Range Photogrammetry and 
Surveying Symposium, ASP, pp. 35- 
46. 
/3/ Curry, 8., Anderson, Calibration of An Array Camera. 
J.M., Baumrind, S.: Photogrammetric Engineering and 
Remote Sensing, in press. 
/A/ Pavlidis, T. : Algorithms for Graphics and Image 
Processing. 
Computer Science Press, 1982. 
/5/ Rosenfeld, A, : Image Analysis : Problems, 
Progress, and Prospects. 
2.175 -