I'm currently working on a piece of code which prepares an image for thresholding, although there seems to be a mismatch between the output image and the pixel data which is returned.
I'll talk you through the code. First, I define the function which takes three parameters - IMAGE, the image that will be thresholded; r the radius which defines a neighbourhood around a central pixel, from which pixel data can be calculated; and k, the 'kernel' which determines the detail of the thresholding.
function [t] = imadaptive ( IMAGE, r, k )
I then call functions I've previously defined, which create images in which the mean and standard deviation of each pixel is stored:
meanpic = immean (IMAGE, r);
stdpic = imstd (IMAGE, r);
Next, I create a new image t, where each pixel has an intensity of 0. This will be used to store the appropriate thresholding values for each pixel:
t = zeros(size(IMAGE));
I then work out the size of the image to tell the for-loop when to stop:
[nx, ny] = size(IMAGE);
Next, I start the for-loop and run a series of if-statements to stop the program from trying to inspect pixel values in positions like (-2,-2):
if x-r <= 0
startx = 1;
else
startx = x-r;
end
if x+r > nx
endx = nx;
else
endx = x+r;
end
if y-r <= 0
starty = 1;
else
starty = y-r;
end
if y+r > ny
endy = ny;
else
endy = y+r;
end
Finally, I run the code to work out the work out the appropriate threshold values for each pixel and then add that value to the image t:
R = 128;
for xp = startx : endx
for yp = starty : endy
if (x-xp)^2 + (y-yp)^2 <= r^2
thresh = meanpic(xp,yp) * (1 + (k * (((stdpic(xp,yp) / R) - 1))));
end
end
end
t(x,y) = thresh;
The problem is: when I run this code, I get an image identical to my stdpic: the picture which shows the standard deviation values of each image. I can't figure out why this is, though. The equation seems to be correct - and when I remove the ; from the equation, I see that the values it outputs are very varied and between 0 and 255, so the output image t should look wildly different.
Can anyone point out what I've got wrong here?
The thresh is reassigned each time through the loop. Thus only one pixel in the neighborhood contributes to each output pixel. If changing thresh doesn't help, would you please post the whole file here or on pastebin? The walkthrough format is very useful, but it's always possible the bug is in the code not shown.
Edit: at lines 163-165 of the paste, change uint8to double. At line 211, add code to use a pixel count or something similar. I think the division by R is happening in uint and truncating to zero. Change back to uint at the very end, after the math, or else imshow won't give you the results you expect. Hope this helps!
Related
I have a clean image, and a noisy image. I created a denoiser and applied it to the noisy image, that was my final output. Now to compare how much this image is close to a clean image I need to compare it using PSNR and SSIM, but due to different positions of the image I am unable to compare.
Now I am getting SSIM as 0.5, which is very low, due to the improper placement of both the images. If the images are registered properly, then I guess SSIM should come around 0.80+. But I have not been able to accomplish this.
How can I align these two images to obtain a good SSIM value?
I have two coin images, 1st image (CLEAN), 2nd image (IMPROVED a NOISY IMG), for comparison.
Clean Img:
Noisy Img:
Due to positions of images at different positions ssim(img1,img2) is giving incorrect output. I tried cropping but that did not work.
Here is what I have tried so far:
Attempt 1:
function [valPSNR,valSSIM,badpict]=getSSIM(clean_img,img2)
% pad reference image since object is so close to edges
refpict = padarray(mat2gray(clean_img),[20 20],'replicate','both');
% crop test image down to extract the object alone
badpict = imcrop(mat2gray(img2),[2.5 61.5 357 363]);
% maximize normalized cross-correlation to find offset
szb = size(badpict);
c = normxcorr2(badpict,refpict);
[idxy idxx] = find(c == max(c(:)));
osy = idxy-szb(1);
osx = idxx-szb(2);
% crop the reference pict to the ROI
refpict = refpict(osy:idxy-1,osx:idxx-1);
%imshow(imfuse(badpict,refpict,'checkerboard'));
%imagesc(badpict);
valSSIM=ssim(badpict,refpict);
valPSNR=getPSNR(badpict,refpict);
img2=badpict;
clean_img=refpict;
figure; imshowpair(clean_img,img2);
figure; montage({mat2gray(clean_img),mat2gray(img2)}, 'Size', [1 2], 'BackgroundColor', 'w', 'BorderSize', [2 2]);
end
Attempt 2:
function [valPSNR,valSSIM,badpict]=getSSIM2(clean_img,img2)
% pad reference image since object is so close to edges
bw1 = im2bw(mat2gray(clean_img));
bw2 = imclose(im2bw(mat2gray(img2),0.3),strel('disk',9));
bw2 = bwareafilt(bw2,1);
% make same size
[r,c] = find(bw1);
clean_img = clean_img(min(r):max(r),min(c):max(c));
[r,c] = find(bw2);
img2 = img2(min(r):max(r),min(c):max(c));
img2= imresize(img2, size(clean_img),'bilinear');
valPSNR=getPSNR(mat2gray(clean_img),mat2gray(img2));
valSSIM=ssim(mat2gray(clean_img),mat2gray(img2));
badpict=img2;
figure; imshowpair(clean_img,img2);
figure; montage({mat2gray(clean_img),mat2gray(img2)}, 'Size', [1 2], 'BackgroundColor', 'w', 'BorderSize', [2 2]);
end
As others have pointed out, the resampling required by registration will have some non-zero error. But, here is some sample code that will take you through the registration part that is the crux of your question.
% SSIM isn't defined on RGB images, convert to grayscale.
ref = rgb2gray(imread('https://i.stack.imgur.com/tPKEJ.png'));
X = rgb2gray(imread('https://i.stack.imgur.com/KmU4y.png'));
% The input image data has bright borders at the edges that create
% artifacts in resampling, best to just crop those or maybe there are
% aquisitions that don't have these borders?
X = X(3:end-2,3:end-2);
ref = ref(4:end-3,4:end-3);
figure
montage({X,ref});
tform = imregcorr(X,ref,"translation");
Xreg = imwarp(X,tform,OutputView=imref2d(size(ref)),SmoothEdges=true);
figure
imshowpair(Xreg,ref)
ssim(Xreg,ref)
Maybe you can refer to my github.
I implemented a template matching algorithm by OpenCV which you can use NCC-Based Pattern Matching to find targets, and then get a score (similarity).
You can then use this score to decide if it is clean.
Besides, tranforming c++ code may be an issue for you, but just find the all corresponded function in matlab version.
Here are effects (red blocks are areas with similarity higher than threshold 0.85 in comparison with golden sample):
The whole function is too long to be posted here.
Part of the function:
for (int i = 0; i < iSize; i++)
{
Mat matRotatedSrc, matR = getRotationMatrix2D (ptCenter, vecAngles[i], 1);
Mat matResult;
Point ptMaxLoc;
double dValue, dMaxVal;
double dRotate = clock ();
Size sizeBest = GetBestRotationSize (vecMatSrcPyr[iTopLayer].size (), pTemplData->vecPyramid[iTopLayer].size (), vecAngles[i]);
float fTranslationX = (sizeBest.width - 1) / 2.0f - ptCenter.x;
float fTranslationY = (sizeBest.height - 1) / 2.0f - ptCenter.y;
matR.at<double> (0, 2) += fTranslationX;
matR.at<double> (1, 2) += fTranslationY;
warpAffine (vecMatSrcPyr[iTopLayer], matRotatedSrc, matR, sizeBest);
MatchTemplate (matRotatedSrc, pTemplData, matResult, iTopLayer);
minMaxLoc (matResult, 0, &dMaxVal, 0, &ptMaxLoc);
vecMatchParameter[i * (m_iMaxPos + MATCH_CANDIDATE_NUM)] = s_MatchParameter (Point2f (ptMaxLoc.x - fTranslationX, ptMaxLoc.y - fTranslationY), dMaxVal, vecAngles[i]);
for (int j = 0; j < m_iMaxPos + MATCH_CANDIDATE_NUM - 1; j++)
{
ptMaxLoc = GetNextMaxLoc (matResult, ptMaxLoc, -1, pTemplData->vecPyramid[iTopLayer].cols, pTemplData->vecPyramid[iTopLayer].rows, dValue, m_dMaxOverlap);
vecMatchParameter[i * (m_iMaxPos + MATCH_CANDIDATE_NUM) + j + 1] = s_MatchParameter (Point2f (ptMaxLoc.x - fTranslationX, ptMaxLoc.y - fTranslationY), dValue, vecAngles[i]);
}
}
FilterWithScore (&vecMatchParameter, m_dScore-0.05*iTopLayer);
I'm relatively new to Matlab, and trying to understand why a piece of code isn't working.
I have a 512x512 image that needs to be downsized to 256, and then resized back up to 512.
How I understand the mathematics, is that I would need to mean the pixels in the image to get the 256, and then sum them back to get the 512. Is that correct ? Following is the code that I'm looking at, and if someone can explain me whats wrong(its giving a blank white image), I would appreciate it:
w = double(imread('walkbridge.tif'));
%read the image
w = w(:,:,1);
for x = 1:256
for y = 1:256
s256(x,y) = (w(2*x,2*y)+ w(2*x,(2*y)-1) + w((2*x)-1,2*y)+ w((2*x)-1,(2*y)-1))/4;
end
end
for x = 1 : 256
for y = 1 : 256
for x1 = 0:1
for y1 = 0:1
R1((2*x)-x1,((2*y)-y1)) = s256(x,y);
end
end
end
end
imshow(R1)
I got your code to work, so you might have some bad values in your image data. Namely, if your image has values in range 0..127 or something similar, it will most likely show as all white. By default, imshow expects color channels to be in range 0..1.
You might also want to simplify your code a bit by indexing the original array instead of accessing individual elements. That way the code is even easy to change:
half_size = 256;
w = magic(2*half_size);
w = w / max(w(:));
figure()
imshow(w)
s = zeros(half_size);
for x = 1:half_size
for y = 1:half_size
ix = w(2*x-1:2*x, 2*y-1:2*y);
s(x,y) = sum(ix(:))/4;
end
end
for x = 1:half_size
for y = 1:half_size
R1(2*x-1:2*x, 2*y-1:2*y) = s(x,y);
end
end
figure()
imshow(R1)
I imagine the calculations could even be vectorised in some way instead of looping, but I didn't bother.
I have a collection of data that I am trying to graph as a histogram. Additionally, I would like to color the individual bars as a function of the x axis location. CData, described here seems to do what I want but I can't get it to work.
Here is my code:
h = bar(new_edge,N,'hist','FaceColor','flat');
hold on
for n = 1:length(N)
if (x - x/1.09) - (x-1) > 0
probability(n) = 1 - ((x-x/1.09) - (x-1))/((x - 1/1.09)+(x/0.91 - x));
else
probability(n) = 1;
end
color_num = 30;
cm = jet(color_num);
min = 0.5450;
max = 1;
color_map_index = floor(1 + (probability(n) - min)/(max-min)*(color_num-1));
rbg = cm(color_map_index,:);
h.CData(n,:) = rbg;
end
Similar to the MATLAB example, I first create my bar graph. Next, I want to loop through and prescribe the color for each bar based on a calculation. I do this by creating a colormap with # of bins and a min/max, getting a color index, then finally retrieving the rbg value. I get the following error when I try to apply the color:
Subscript indices must either be real positive integers or logicals.
h.CData(n,:) = rbg;
If I dive into the h object, MATLAB tells me that CData has a size of (4x65). What's going on here? Both new_edge and N are 1x65 vectors.
Are you certain that the error you're getting ("Subscript indices must either be real positive integers or logicals") is coming from the following line?:
h.CData(n,:) = rbg;
Since we know n is a positive integer greater than or equal to one, the indexing here shouldn't have a problem. It's more likely that your error is coming from the line above it (i.e. the value for color_map_index is less than 1). I would double-check how you are computing color_map_index.
You could also try using function notation (i.e. get and set) instead of dot notation to update the property:
cData = get(h, 'CData');
cData(n, :) = rbg;
set(h, 'CData', cData);
Incidentally, you should also not be giving your variables the same name as existing functions, like you are doing here:
...
min = 0.5450;
max = 1;
...
This shadows the built-in min and max functions, which can also lead to the same error message under other conditions. Definitely rename those.
If you are still having trouble after trying those fixes, you could try setting the color using indexed color mapping instead, as illustrated in one of my other answers (near the bottom). As a simple example, the following plots 20 bars of 30 different possible values, then colors them based on their height:
color_num = 30;
N = randi(color_num, 1, 20);
hBar = bar(N, 'hist');
colormap(parula(color_num));
set(hBar, 'CData', N, 'CDataMapping', 'direct');
And the plot:
This could be a problem with your Matlab version.
When I test CData with bar on 2017b, this works:
openExample('graphics/ControlIndividualBarColorsExample')
When I try it on 2017a, it doesn't run.
Does the example work?
Given that this is a version control problem, there's really not a clean solution. In case anyone else comes along with a similar question and has the same version, here's a workaround that worked for me in 2017a.
Rather than creating a bar chart, you can simply draw rectangles. It's messy, but it does produce the desired result.
[N,edges] = histcounts(AB(AB<2));
probability = zeros(1,length(N));
new_edge = zeros(1,length(N));
for m = 1:length(N)
new_edge(m) = (edges(m) + edges(m+1))/2;
end
figure
hold on
for n = 1:length(N)
x = new_edge(n);
if ((x - x/1.09) - (x-1)) > 0
probability(n) = 1 - ((x-x/1.09) - (x-1))./((x - x/1.09)+(x/0.91 - x));
else
probability(n) = 1;
end
color_num = 100;
cm = jet(color_num);
min = 0;
max = 1;
color_map_index(n) = floor(1 + (probability(n) - min)/(max-min)*(color_num-1));
rbg = cm(color_map_index(n),:);
rectangle('Position',[edges(n),0,(edges(n+1)-edges(n)),N(n)],'Edgecolor','k','FaceColor',rbg)
end
set(gcf,'color','w');
blah = colorbar;
I have edited a code that i found online that helps me draw points somehow distributed on a graph based on the minimum distance between them
This is the code that i have so far
x(1)=rand(1)*1000; %Random coordinates of the first point
y(1)=rand(1)*1000;
minAllowableDistance = 30; %IF THIS IS TOO BIG, THE LOOP DOES NOT END
numberOfPoints = 300; % Number of points equivalent to the number of sites
keeperX = x(1); % Initialize first point
keeperY = y(1);
counter = 2;
for k = 2 : numberOfPoints %Dropping another point, and checking if it can be positioned
done=0;
trial_counter=1;
while (done~=1)
x(k)=rand(1)*1000;
y(k)=rand(1)*1000;
thisX = x(k); % Get a trial point.
thisY = y(k);
% See how far is is away from existing keeper points.
distances = sqrt((thisX-keeperX).^2 + (thisY - keeperY).^2);
minDistance = min(distances);
if minDistance >= minAllowableDistance
keeperX(k) = thisX;
keeperY(k) = thisY;
done=1;
trial_counter=trial_counter+1;
counter = counter + 1;
end
if (trial_counter>2)
done=1;
end
end
end
end
So this code is working fine, but sometimes matlab is freezing if the points are above 600. The problem is full , and no more points are added so matlab is doing the work over and over. So i need to find a way when the trial_counter is larger than 2, for the point to find a space that is empty and settle there.
The trial_counter is used to drop a point if it doesn't fit on the third time.
Thank you
Since trial_counter=trial_counter+1; is only called inside if minDistance >= minAllowableDistance, you will easily enter an infinite loop if minDistance < minAllowableDistance (e.g. if your existing points are quite closely packed).
How you do this depends on what your limitations are, but if you're looking at integer points in a set range, one possibility is to keep the points as a binary image, and use bwdist to work out the distance transform, then pick an acceptable point. So each iteration would be (where BW is your stored "image"/2D binary matrix where 1 is the selected points):
D = bwdist(BW);
maybe_points = find(D>minAllowableDistance); % list of possible locations
n = randi(length(maybe_points)); % pick one location
BW(maybe_points(n))=1; % add it to your matrix
(then add some checking such that if you can't find any allowable points the loop quits)
I have different images and I would like to crop them and keep only what is different between both. Here is the code I have so far.
video = VideoReader('frames.avi', 'Tag', 'my reader object');
frameFirst = read(video,1);
frameSecond = read(video, video.NumberOfFrames-1 );
imshowpair (frameSecond,frameFirst);
pause();
Well, it's tough to give you a good answer without much more detail. I think I understand what you're trying to do, and this might get you moving in the right direction. This code iterates through each pixel of the image (each pixel contains a 1x3 vector of RGB data ranging from 0 to 1), by row and column. If the difference in any of the elements of the 1x3 RGB vector exceeds some threshold (in this case, set to 0.1), we make that whole pixel black (set it to [0 0 0]). Else, we lust make it whatever the last frame was. To filter out all but those pixels that are identical, set the thresh value to 0. It goes like this:
thresh = 0.1
for ii = 1:size(frameFirst, 1)
for jj = 1:size(frameFirst, 2)
pixDiff = frameFirst{ii, jj} - frameSecond{ii, jj}
if (pixDiff(1) > thresh || pixDiff(2) > thresh || pixDiff(3) > thresh)
outputFrame = frameSecond{ii, jj};
else
outputFrame = [0 0 0];
end
end
end
I hope this does what you're looking for. Good luck!
Edit 1: Ok, I understand what you are looking for now. You need to have the indices of the bottom-right and top-left. If you already have those, just do this: frameOut = frameIn(xStart:xStop, yStart, yStop. If you need to find those points, that's harder. Let me know and I'll help you work it out.