How to divide a curve into 8 equal region along x axis?

Hi. I am working with border irregularity. For this I have determined the irregularities in the curve. Here the 'stars' indicate the irregularities. I want to divide it into 8 parts so that if any part contains minimum one star I can count it as 1. So if all the 8 parts have stars in each,the value will be 8. So the final value will be of any between 0 to 8. How can I do this?

I have attached my code and the image.

[KSSV]

How about using reshape?

[joynob ahmed]

But when I used reshape it didn't give me the distribution of star in that way

Answer 1

[KSSV]

A = rand(1,100) ; 
a = length(A) ; 
n = 8 ; 
% check whether data needed to be appended to reshape into n 
b = n * ceil(a / n)-a ;
% 
A = [A NaN(1,b)] ;
iwant = reshape(A,n,[]) ; 
% plot 
plot(iwant)

[joynob ahmed]

I didn't understand it. How can this divide it into 8 part? I wanted to count if the part has the star so that I can count them.

[KSSV]

Star is nothing but zeros of the curve. Did you try that first place?

[joynob ahmed]

When I used this:

n = 8 ;

a=length(j);

% check whether data needed to be appended to reshape into n

b = n * ceil(a / n)-a

%

j = [j NaN(1,b)] ;

iwant = reshape(j,n,[]) ;

% plot

figure

plot(iwant)

I got this error:

Dimensions of arrays being concatenated are not consistent

in this line: j=[j NaN(1,b)];

[KSSV]

It looks like you have a column vector...transpose your vector to row vector.

[joynob ahmed]

I am sorry. I am not good at coding. I can't understand your code. But I have written this:

n = 8 ; 
d=j';
a=length(j);
% check whether data needed to be appended to reshape into n 
b = n * ceil(a / n)-a ;
% 
d = [d NaN(1,b)] ;
iwant = reshape(d,n,[]) ; 
% plot 
figure
plot(iwant)

And got this:

May be I didn't explain well. please see this and help me out: https://www.mathworks.com/matlabcentral/answers/551446-how-to-split-a-curve-into-8-equal-parts#comment_906826

[KSSV]

What is size of iwant?

[joynob ahmed]

8 by 497

[KSSV]

This is what you wanted right???

[joynob ahmed]

what I wanted is like this:

You can see among these 8 regions,7 regions got minimum 1 star in them.So the value assigned for this lesion will be 7.

[joynob ahmed]

can you help me please?

[KSSV]

What I gave code for does the same.

[joynob ahmed]

But this code reshapes such that the first 8 point fall into 8 different region where as I wanted to divide so that 1st 497 points which means all the point in first region to be in the x=1.I have checked the result.If I do so the result won't be correct.

Is there any way to rearrange this?

[joynob ahmed]

It works when I changed the row and column of the matrix. Thank you for your answer.

Answer 2

[John D'Errico]

I am surely wasting my time (and yours) to answer this, because I don't understand the question. Far too confusingly posed to make sense. I think you are asking how to break the curve into exactly 8 parts, each as equal in length as possible. I can only assume you mean distance traveled along the curve, so arclength. Sadly, arclength is not terribly well defined on a noisy curve, but that is something only you can deal with. I'd assume just a piecewise linear arclength, so a connect-the-dots curve.

Next, you insist that one of those stars MUST appear in each segment. The starred points "seem" to be zero-crossings, but even that is not clear. Are the stars every zero crossing of this noisy curve? How did you choose those points? Can a starred point be shared across two segments? That is, if a starred point lies at the break point between two segments, is that sufficient?

Next, what is the point of this task? What is your goal?

I ask that question because far too often I see people ask a specific confusing question on Answers, when their real goal is something relatively simple to solve, IF they knew how to solve that problem. So, is your goal in this task to locate those roots/zero crossings?

If you really wanted to break the curve into segments that are all roughly as much as possible equal in length, such that each segment MUST include at least one of those starred points. Then...

Find the total arc length along the curve.

Break the curve into 8 segments, where each segment has the same arc length. I have posted a tool on the file exchange which can do exactly this. Test to see if each segment contains at least one starred point. If not, then adjust the segment lengths, perhaps using an optimization tool, until the goal is satisfied.

If your final goal is something different, so as I suggested, it is merely to find the zero crossings of that function, there are tools that can do this trivially in one call.

[joynob ahmed]

I think I didn't explain so well.

I have using this code for finding border irregularity of a lesion.For this I have finished upto the determination of border irregularity function and pointing the irregularity. The ''star'' indicates the irregularity. I have done upto this. Now I want to split the graph along x axis into 8 equal parts. Then I want to see if each splitted region/part got minimum 1 abrupt cutoff. If there is an abrupt cutoff then the score of the region will be 1. Thus counting them, the value can be any between 0-8.And according to this I will take decision. I think you can understand my problem.

[Image Analyst]

So is the minimum for a 1/8th segment of x one star or zero stars?

And I do not understand the grammar of this sentence: "Then I want to see if each splitted region/part got minimum 1 abrupt cutoff." It seems there are some crucial words missing in that sentence.

[joynob ahmed]

You can see this comment: https://www.mathworks.com/matlabcentral/answers/551446-how-to-divide-a-curve-into-8-equal-region-along-x-axis#comment_909298

when a region got star,its value will be 1. Abrupt cut off indicates star here.