MATLAB: I have an image which is full black background and the image contain few of white point. How can i draw a straight line that connected most of the white point as shown in figure.

straight linetrend line

I cant find a way to draw a straight line which on a black background that contain few white point on the image.

Best Answer

To avoid confusion, this answer is based on correspondence that has since been deleted. Original poster is looking for a type of upper bound trend line that is fitted to final part of the data set. Other answers show how to connect the dots (Image Analyst) and how to perform linear regression (Rik Wisselink).

The attached script uses ischange to subdivide the data set and fit linear segments. You can experiment with the MaxNumChanges and Threshold arguments of ischange to change the nature of the line.

Related Solutions

MATLAB: How to connect these components in the image to make the crack continuous？

Try this, it's quite robust. However, you may want to exclude 'outliers' before interpolating.

%%Read image
I=imread('kAreaOp14.png');
%%Find coordiantes of crack
[y,x]=find(I==1);
%%Linear Interpolation
[x, index] = unique(x); 
xi=min(x):max(x);
yi = interp1(x, y(index), xi) 
%%round to integer
yi=round(yi);
xi=round(xi);
%%Convert coords to linear index
inds=sub2ind(size(I),yi,xi)
%%create empty image
out=zeros(size(I),'logical')
%%add interpolated line
out(inds)=1;
%%superimpose original data and plot
out=out+I;
out(out>1)=1;
imshow(out)

MATLAB: How to use fit in logY

A line which is straight in a linear plot will always be curved in a log plot...? If you want it to be straight in a log10 plot, your original fit has to be exponential 10^x.

What you can do is e.g. get a linear fit g(x) to log10(y), and then f(x) = 10.^g(x) will be your actual regression function (which is obviously exponential).

Best Answer

Related Solutions

MATLAB: How to connect these components in the image to make the crack continuous？

MATLAB: How to use fit in logY

Related Question