A very interesting approach to the integration of several features into the 
process of analyzing shape may be based on [1, Theorem 2, Section 3.5] by 
Aloimonos and Shulman . In this study, the implementation of this approach was 
tested by comparing the results of three different algorithms. For applying this 
approach, several specific model assumptions and situations in scene space have 
to be carefully considered. A dependence could be proved of the accuracy of 
shape results from the direction to the point light source, and from the 
complexity of the realized motion in 3-D space.