Find Threshold on Canny edge image problem

Dear all experts,

I met a question about how to find the threshold values by giving a fixed edge percentage on canny edge detection problem. I use two 5x5 gradient filters. The code is showns below:

% convolve filterX and filterY with Image to get magnitude gradient and angle
Ix = conv2(Image,filterX); % filterX is a 5x5 gradient filter
Iy = conv2(Image,filterY); % filterY is a 5x5 gradient filter
magnitude = sqrt(Ix.^2+Iy.^2);
angle = atan2(Iy,Ix);

% How to find threshold based on the %percentage? The edge_percentage is the 
% proportion of pixels supposed to be treated as potential edge points 
max = max(max(magnitude));
min = min(min(magnitude));
% threshold = some percentage work

[Image Analyst]

We have no idea what percentage of your objects you expect to be edge points. It depends on the image. Obviously a circle has the lowest proportion of edge points, while an asterisk * shaped blob will have a much higher percentage.

[Tony123]

the value of the percentage is set as 90% as default

[Image Analyst]

Again, it depends on the image. And you can set a threshold to get any binary image you want, but percentage of What? What is your reference? Attach your image.

[Tony123]

I attached the target picture

[KALYAN ACHARJYA]

Sorry I am not getting the question.

[Tony123]

It's to find a threshold such that a fixed percentage of the gradients are considered to be possible edge points. Then I will use the calculated edge orientation, create 4 binary images representing the horizontal, vertical, diagonally up and diagonally down edges exceeding the threshold. I also attached the gradient filters I am going to use

[Image Analyst]

Make it easy for us to help you. Give us code to create filterX and filterY.

[Tony123]

Sorry for the unclear, I attached my code