DIRECT CONSTRUCTION OF POLYNOMIAL SURFACES FROM DENSE RANGE IMAGES THROUGH REGION GROWING

Existing techniques for approximating discrete data with smooth surfaces are based on either curve fitting or edge detection. Furthermore, they often rely on the user's expertise for determining the number of polynomial segments included in each surface. This paper describes a technique for approximating 3-D points with a functional surface z = z(x, y) based on the concept of ''region growing,'' aimed at maximizing the amount of data represented by a single polynomial segment. The resulting methodology requires minimal user interaction and makes full use of the ''surface information'' existing in a very dense set of digitized points produced by today's highly accurate range imaging sensors. Experimental results are presented showing the algorithm's performance on a variety of data.