Why is NURBS prefered over B-Spline in CAD software? Why is NURBS prefered over B-Spline in CAD software? That too NURBS is exhaustively used in solid modelling packages like Pro-E, CATIA etc., What are the advantages of NURBS over B-Spline. 
Why is it easier to parametrically regenerate NURBS surfaces?
 A: Actually, the main reason that NURBs are popular in CAD systems is because they're popular in CAD systems :-) Seriously -- every CAD system today has to implement NURBs because all the other systems have them, and data exchange between systems is crucially important.
It wasn't always this way, of course. The NURBs movement got started when they were introduced into the IGES standard in 1981. This was mostly precipitated by Boeing. It was clear that something new was needed because the existing free-form geometry in IGES was pretty poor, so there wasn't much resistance. The NURBS sales pitch was (more-or-less verbatim):
(1) They are invariant under affine transformations
(2) Operations like rotations and translations can be applied to NURBS curves and surfaces by applying them to their control points.
(3) They offer one common mathematical form for both standard analytical shapes 
(e.g., conics) and free-form shapes.
(4) They provide the flexibility to design a large variety of shapes.
(5) They reduce the memory consumption when storing shapes (compared to simpler methods).
(6) They can be evaluated and differentiated reasonably quickly by numerically stable and accurate algorithms.
I think several of these reasons are misleading half-truths. I'll comment on a few of them as examples:
Item #3:  Able to represent circles and conics. This isn't quite true. NURBs can represent the shapes of circles and conics, but not their conventional trigonometric parameterizations. And parameterizations are important in CAD. Also, CAD systems already had good methods for representing circles, so there was no reason to introduce another (inferior) method.
Item #5: Reduced storage space. It's true that NURBs occupy less space than the spline forms previously used in IGES, but they are much larger than analytic representations of cylinders, spheres, tori, etc.
Item #6: Easy to evaluate and differentiate. Well, they're quite a bit harder than polynomial splines. And what about when you want to integrate??
But these are quibbles, really -- having some standard is the most important point. It doesn't matter that it's not quite the right one, or that the reasons for choosing it are a bit shakey.
You might be interested in reading this paper and some of its references.
In this paper, some of the original proponents of NURBS explain why they might have made a mistake.
A: NURBS are a generalization of B-Splines.
Here is one difference between them.
To apply an affine transformation (e.g. scaling, rotation and translation) to a B-Spline one can apply the transformation on the control points and the transformed B-Spline will then be represented by the new control points. For NURBS this is not only true for affine transformations but also for projective transformations.
A: You will find quite nice explanations looking at
http://en.wikipedia.org/wiki/Non-uniform_rational_B-spline
A: One reason is that NURBS can represent conics exactly, and circles are important in CAD.
