# Get smoothing point using B-spline curve C++

I'm working on video stabilization topic. At the smoothing part, I need smooth parameters (translation, rotation+scale) by using B-spline curve for warping in order to create stabilized video.

Now, I am testing on some points. For example, I have 4 points (control points) with degree = 2, after using b-spline I wanna obtain 4 smoothed points. But when I use B-spline curve sample code in below, it created more than 4 points. I can not understand exactly that those results points which are belong to original points (control points). If I only get 4 points, the results is like as 4 original points (4 control points).How to get 4 smoothed points by using B-Spline curve?

Hope you guys help me to show this problem.

Here is the code.

#include "stdafx.h"
#include <stdio.h>
#include <conio.h>
/*
Subroutine to generate a B-spline open knot vector with multiplicity
equal to the order at the ends.

c            = order of the basis function
n            = the number of defining polygon vertices
nplus2       = index of x() for the first occurence of the maximum knot vector value
nplusc       = maximum value of the knot vector -- $n + c$
x()          = array containing the knot vector
*/
void knot(int n, int c, int x[])
{
int nplusc,nplus2,i;

nplusc = n + c;
nplus2 = n + 2;

x[1] = 0;
for (i = 2; i <= nplusc; i++){
if ( (i > c) && (i < nplus2) )
x[i] = x[i-1] + 1;
else
x[i] = x[i-1];
}
}
/*  Subroutine to generate B-spline basis functions for open knot vectors

C code for An Introduction to NURBS
by David F. Rogers. Copyright (C) 2000 David F. Rogers,

Name: basis.c
Language: C
Subroutines called: none
Book reference: p. 279

c        = order of the B-spline basis function
d        = first term of the basis function recursion relation
e        = second term of the basis function recursion relation
npts     = number of defining polygon vertices
n[]      = array containing the basis functions
n[1] contains the basis function associated with B1 etc.
nplusc   = constant -- npts + c -- maximum number of knot values
t        = parameter value
temp[]   = temporary array
x[]      = knot vector
*/

void basis(int c,float t, int npts,int x[],float n[])
{
int nplusc;
int i,k;
float d,e;
float temp[36];

nplusc = npts + c;

/* calculate the first order basis functions n[i][1]    */

for (i = 1; i<= nplusc-1; i++){
if (( t >= x[i]) && (t < x[i+1]))
temp[i] = 1;
else
temp[i] = 0;
}

/* calculate the higher order basis functions */

for (k = 2; k <= c; k++){
for (i = 1; i <= nplusc-k; i++){
if (temp[i] != 0)/* if the lower order basis function is zero skip the                         calculation */
d = ((t-x[i])*temp[i])/(x[i+k-1]-x[i]);
else
d = 0;

if (temp[i+1] != 0)     /* if the lower order basis function is zero skip the calculation  */
e = ((x[i+k]-t)*temp[i+1])/(x[i+k]-x[i+1]);
else
e = 0;

temp[i] = d + e;
}
}

if (t == (float)x[nplusc]){     /*    pick up last point    */
temp[npts] = 1;
}

/* put in ...
edit retag close merge delete

Sort by » oldest newest most voted

The p1 = 11; determines the number of evaluated points. But if only fixed number of points needed, its a waste using a generic B-spline. The exact behaviour can be achived with only weighting the four points with precalculated B-spline basis function values. And the k = 2; means that the segments are simple lines which are just connect the control points, so it should be 3 to be continuous in tangent, and 4 to be continuous in curvature.

So quick fix is p1 = 4; and k = 3;

But this is not the best solution to filter out points.

An easy: Moving average A hard one: Kalman filter

more

@kovand11: Thank you so much. I did it. About using B-spline curve to filter out points, that's a way I want to test it.

( 2014-06-08 08:31:36 -0500 )edit

Official site

GitHub

Wiki

Documentation