fitPolynomialRANSAC

Fit polynomial to points using RANSAC

collapse all in page

Syntax

P = fitPolynomialRANSAC(xyPoints,N,maxDistance)
[P,inlierIdx] = fitPolynomialRANSAC(___)
[___] = fitPolynomialRANSAC(___,Name=Value)

Description

P = fitPolynomialRANSAC(xyPoints,N,maxDistance) finds the Nth-degree polynomial coefficients, P, by sampling a small set of points given in xyPoints and generating the Nth polynomial fits. The fit that has the most inliers within maxDistance is returned. If a fit cannot be found, then P is returned empty. The function uses the M-estimator sample consensus (MSAC) algorithm, a variation of the random sample consensus (RANSAC) algorithm to fit the data.

example

[P,inlierIdx] = fitPolynomialRANSAC(___) returns a logical array, inlierIdx, that specifies the indices for data points that are inliers to the fit polynomial based on maxDistance. Use the input arguments from the previous syntax.

[___] = fitPolynomialRANSAC(___,Name=Value) specifies options using one or more name-value arguments in addition to any combination of arguments from previous syntaxes. For example, (MaxNumTrials=2000) sets the maximum number of random trials to 2000.

Examples

collapse all

Open Live Script

Use the RANSAC algorithm to generate a polynomial that fits a set of noisy data. The fitPolynomialRANSAC function generates a polynomial by sampling a small set of points from [x y] point data and generating polynomial fits. The fit with the most inliers within maxDistance is returned.

Construct and plot a parabola with [x y] points.

x = (-10:0.1:10)';
y = (36-x.^2)/9;
figure
plot(x,y)
title('Parabola')

Figure contains an axes object. The axes object with title Parabola contains an object of type line.

Add noise and outlier points to the points on the parabola.

y = y+rand(length(y),1);
y([50,150,99,199]) = [y(50)+12,y(150)-12,y(99)+33,y(199)-23];

plot(x,y)
title('Parabola with Outliers and Noise')

Figure contains an axes object. The axes object with title Parabola with Outliers and Noise contains an object of type line.

Use fitPolynomialRANSAC to generate coefficients for a second-degree polynomial. Also get the inliers identified by the specified maxDistance from the polynomial fit.

N = 2;           % second-degree polynomial
maxDistance = 1; % maximum allowed distance for a point to be inlier

[P, inlierIdx] = fitPolynomialRANSAC([x,y],N,maxDistance);

Evaluate the polynomial using polyval. Plot the curve and overlay the [x y] points. Mark outliers with a red circle.

yRecoveredCurve = polyval(P,x);
figure
plot(x,yRecoveredCurve,'-g','LineWidth',3)
hold on
plot(x(inlierIdx),y(inlierIdx),'.',x(~inlierIdx),y(~inlierIdx),'ro')
legend('Fit polynomial','Inlier points','Outlier points')
hold off

Figure contains an axes object. The axes object contains 3 objects of type line. One or more of the lines displays its values using only markers These objects represent Fit polynomial, Inlier points, Outlier points.

Input Arguments

collapse all

[x y] coordinate points, specified as an m-by-2 matrix. The polynomial is fit to these points.

Data Types: double | single | uint32 | int32 | uint16 | int16

Degree of polynomial fit, P, specified as an integer. The degree of a polynomial is the highest degree of the terms in the equation. For example, a polynomial of degree 2 is:

Ax2+Bx+C

A, B, and C are constants. In general, higher degree polynomials allow for a better fit, but the fit depends on your data.

Maximum distance from the polynomial fit curve to an inlier point, specified as a positive scalar. Any points further away are considered outliers. The RANSAC algorithm creates a fit from a small sample of points but tries to maximize the number of inlier points. Lowering the maximum distance helps to improve the polynomial fit by putting a tighter tolerance on inlier points. For an M-by-2 matrix of points in the form [x y], the distance is defined as abs(y - polyval(model,x)).

Name-Value Arguments

collapse all

Specify optional pairs of arguments as Name1=Value1,...,NameN=ValueN, where Name is the argument name and Value is the corresponding value. Name-value arguments must appear after other arguments, but the order of the pairs does not matter.

Before R2021a, use commas to separate each name and value, and enclose Name in quotes.

Example: (MaxNumTrials=2000) sets the maximum number of random trials to 2000.

Maximum number of random trials, specified as an integer. A single trial uses a minimum number of random points from xyPoints to fit a parabolic model. Then, the trial checks the number of inliers within the maxDistance from the model. After all trials, the model with the highest number of inliers is selected. Increasing the number of trials improves the robustness of the output at the expense of additional computation.

Confidence that the final solution finds the maximum number of inliers for the polynomial fit, specified as a scalar from 0 to 100. Increasing this value improves the robustness of the output at the expense of additional computation.

Function to validate polynomial, specified as a function handle. The function returns true if the polynomial is accepted based on criteria defined in the function. Use this function to reject specific polynomial fits. The function must be of the form:

isValid = validatePolynomialFcn(P,varargin)

If no function is specified, all polynomials are assumed to be valid.

Maximum number of attempts to find a sample that yields a valid polynomial, specified as an integer.

Output Arguments

collapse all

Polynomial coefficients, returned as a vector of numeric scalars. Each element corresponds to a constant in the polynomial equation with degree N. For example, for a second-degree polynomial, Ax2+Bx+C:

P = [A B C];

Data Types: single | double

Inlier points, returned as a logical vector. The vector is the same length as xyPoints, and each element indicates if that point is an inlier for the polynomial fit based on maxDistance.

References

[1] Torr, P. H. S., and A. Zisserman. "MLESAC: A New Robust Estimator with Application to Estimating Image Geometry." Computer Vision and Image Understanding. Vol. 18, Issue 1, April 2000, pp. 138–156.

Version History

Introduced in R2017a

See Also

| |