Massive sets of data points (X,Y,Z) are generated in aerial mapping, hydrographic surveying and field surveying by Government agencies and private industry. The most straightforward way of thinning large data sets would be to arbitrarily discard every other point for a factor of two size reduction, or you could discard every fourth point, etc. Unfortunately, this means that valuable points of interest are being discarded along with unncessary points. It would certainly be more desirable to eliminate only those points which are redundant in that they fall on a surface that can be delineated with less points while retaining all of the accuracy of the original surface.

Our intelligent data thinning program IntelliThin works by first fast-triangulating the entire set of point into a triangle network. Then, the program makes consecutive passes at thinning with an iteratively greater accuracy loss to generate a point reduction vs. accuracy loss plot, from which an operator may judge the tradeoff between reduction in points and loss of accuracy and perform the final thinning at a specified accuracy loss. The actual thinning is performed by processing every vertex in the triangle mesh and identifying those which are NOT responsible for a single contour crossing in ANY of the triangles containing the vertex. By only eliminating these vertices, the program ensures that areas with varying depths are not thinned while point density in relatively flat areas is greatly reduced.