I am trying to implement the 2D correlation algorithm to detect the position of an object in the image, i don't want to use any built in function estimates 2d correlation.
Here is my code:
I=imread('image.tif'); % image is a black image contains white letters.
h=imread('template.tif'); %template is a small image taken from the original image, it contains one white letter.
I=double(I);
h=double(h);
[nrows ncolumns]=size(I);
[nrows2 ncolumns2]=size(h);
C=zeros(nrows,ncolumns);
for u=1:(nrows-nrows2+1)
for v=1:(ncolumns-ncolumns2+1)
for x=1:nrows2
for y=1:ncolumns2
C(u,v)=C(u,v)+(h(x,y)*I(u+x-1,v+y-1));
end
end
end
end
[maxC,ind] = max(C(:));
[m,n] = ind2sub(size(C),ind) % the index represents the position of the letter.
output_image=(3.55/4).*C./100000;
imshow(uint8(output_image));
I think it is working! but it is very slow.
How can i replace the following code by a better code to speed up the algorithm?
for x=1:nrows2
for y=1:ncolumns2
C(u,v)=C(u,v)+(h(x,y)*I(u+x-1,v+y-1));
end
end
I am thinking that in every time i have the following two matrices
h(1:nrows2,1:ncolumns2) and I(u:u+nrows2-1,v:v+ncolumns2-1)
another question, are there any improvements?
thanks.
Whenever you can, try to use matrix ops. So try something like:
rowInds = (1:nrows2)-1;
colInds = (1:ncolumns2)-1;
temp = h.*I(u+rowInds,v+colInds);
C(u,v) = sum(temp(:));
Instead of:
for x=1:nrows2
for y=1:ncolumns2
C(u,v)=C(u,v)+(h(x,y)*I(u+x-1,v+y-1));
end
end
Yes there are many improvements. You don't need a for loop at all. Since you do not want to use matlab's xcorr2 function, you can use conv2. See the answer I gave here.
How about determining the cross correlation in the Fourier domain, following the cross-correlation theorem? That should guarantee a dramatic speed increase.
Related
I have this 3D array in MATLAB (V: vertical, H: horizontal, t: time frame)
Figures below represent images obtained using imagesc function after slicing the array in terms of t axis
area in black represents damage area and other area is intact
each frame looks similar but has different amplitude
I am trying to visualize only defect area and get rid of intact area
I tried to use 'threshold' method to get rid of intact area as below
NewSet = zeros(450,450,200);
for kk = 1:200
frame = uwpi(:,:,kk);
STD = std(frame(:));
Mean = mean(frame(:));
for ii = 1:450
for jj =1:450
if frame(ii, jj) > 2*STD+Mean
NewSet(ii, jj, kk) = frame(ii, jj);
else
NewSet(ii, jj, kk) = NaN;
end
end
end
end
However, since each frame has different amplitude, result becomes
Is there any image processing method to get rid of intact area in this case?
Thanks in advance
You're thresholding based on mean and standard deviation, basically assuming your data is normally distributed and looking for outliers. But your model should try to distinguish values around zero (noise) vs higher values. Your data is not normally distributed, mean and standard deviation are not meaningful.
Look up Otsu thresholding (MATLAB IP toolbox has it). It's model does not perfectly match your data, but it might give reasonable results. Like most threshold estimation algorithms, it uses the image's histogram to determine the optimal threshold given some model.
Ideally you'd model the background peak in the histogram. You can find the mode, fit a Gaussian around it, then cut off at 2 sigma. Or you can use the "triangle method", which finds the point along the histogram that is furthest from the line between the upper end of the histogram and the top of the background peak. A little more complex to explain, but trivial to implement. We have this implemented in DIPimage (http://www.diplib.org), M-file code is visible so you can see how it works (look for the function threshold)
Additionally, I'd suggest to get rid of the loops over x and y. You can type frame(frame<threshold) = nan, and then copy the whole frame back into NewSet in one operation.
Do I clearly understand the question, ROI is the dark border and all it surrounds? If so I'd recommend process in 3D using some kind of region-growing technique like watershed or active snakes with markers by imregionalmin. The methods should provide segmentation result even if the border has small holes. Than just copy segmented object to a new 3D array via logic indexing.
UPDATE
Here is my code that is meant to add up the two matrices and using element by element addition and then divide by two.
function [ finish ] = stackAndMeanImage (initFrame, finalFrame)
cd 'C:\Users\Disc-1119\Desktop\Internships\Tracking\Octave\highway\highway (6-13-2014 11-13-41 AM)';
pkg load image;
i = initFrame;
f = finalFrame;
astr = num2str(i);
tmp = imread(astr, 'jpg');
d = f - i
for a = 1:d
a
astr = num2str(i + 1);
read_tmp = imread(astr, 'jpg');
read_tmp = rgb2gray(read_tmp);
tmp = tmp :+ read_tmp;
tmp = tmp / 2;
end
imwrite(tmp, 'meanimage.JPG');
finish = 'done';
end
Here are two example input images
http://imgur.com/5DR1ccS,AWBEI0d#1
And here is one output image
http://imgur.com/aX6b0kj
I am really confused as to what is happening. I have not implemented what the other answers have said yet though.
OLD
I am working on an image processing project where I am now manually choosing images that are 'empty' or only have the background, so that my algorithm can compute the differences and then do some more analysis, I have a simple piece of code that computes the mean of the two images, which I have converted to grayscale matrices, but this only works for two images, because when I find the mean of two, then take this mean and find the mean of this versus the next image, and do this repeatedly, I end up with a washed out white image that is absolutely useless. You can't even see anything.
I found that there is a function in Matlab called imFuse that is able to average images. I was wondering if anyone knew the process that imFuse uses to combine images, I am happy to implement this into Octave, or if anyone knew of or has already written a piece of code that achieves something similiar to this. Again, I am not asking for anyone to write code for me, just wondering what the process for this is and if there are already pre-existing functions out there, which I have not found after my research.
Thanks,
AeroVTP
You should not end up with a washed-out image. Instead, you should end up with an image, which is technically speaking temporally low-pass filtered. What this means is that half of the information content is form the last image, one quarter from the second last image, one eight from the third last image, etc.
Actually, the effect in a moving image is similar to a display with slow response time.
If you are ending up with a white image, you are doing something wrong. nkjt's guess of type challenges is a good one. Another possibility is that you have forgotten to divide by two after summing the two images.
One more thing... If you are doing linear operations (such as averaging) on images, your image intensity scale should be linear. If you just use the RGB values or some grayscale values simply calculated from them, you may get bitten by the nonlinearity of the image. This property is called the gamma correction. (Admittedly, most image processing programs just ignore the problem, as it is not always a big challenge.)
As your project calculates differences of images, you should take this into account. I suggest using linearised floating point values. Unfortunately, the linearisation depends on the source of your image data.
On the other hand, averaging often the most efficient way of reducing noise. So, there you are in the right track assuming the images are similar enough.
However, after having a look at your images, it seems that you may actually want to do something else than to average the image. If I understand your intention correctly, you would like to get rid of the cars in your road cam to give you just the carless background which you could then subtract from the image to get the cars.
If that is what you want to do, you should consider using a median filter instead of averaging. What this means is that you take for example 11 consecutive frames. Then for each pixel you have 11 different values. Now you order (sort) these values and take the middle (6th) one as the background pixel value.
If your road is empty most of the time (at least 6 frames of 11), then the 6th sample will represent the road regardless of the colour of the cars passing your camera.
If you have an empty road, the result from the median filtering is close to averaging. (Averaging is better with Gaussian white noise, but the difference is not very big.) But your averaging will be affected by white or black cars, whereas median filtering is not.
The problem with median filtering is that it is computationally intensive. I am very sorry I speak very broken and ancient Octave, so I cannot give you any useful code. In MatLab or PyLab you would stack, say, 11 images to a M x N x 11 array, and then use a single median command along the depth axis. (When I say intensive, I do not mean it couldn't be done in real time with your data. It can, but it is much more complicated than averaging.)
If you have really a lot of traffic, the road is visible behind the cars less than half of the time. Then the median trick will fail. You will need to take more samples and then find the most typical value, because it is likely to be the road (unless all cars have similar colours). There it will help a lot to use the colour image, as cars look more different from each other in RGB or HSV than in grayscale.
Unfortunately, if you need to resort to this type of processing, the path is slightly slippery and rocky. Average is very easy and fast, median is easy (but not that fast), but then things tend to get rather complicated.
Another BTW came into my mind. If you want to have a rolling average, there is a very simple and effective way to calculate it with an arbitrary length (arbitrary number of frames to average):
# N is the number of images to average
# P[i] are the input frames
# S is a sum accumulator (sum of N frames)
# calculate the sum of the first N frames
S <- 0
I <- 0
while I < N
S <- S + P[I]
I <- I + 1
# save_img() saves an averaged image
while there are images to process
save_img(S / N)
S <- -P[I-N] + S + P[I]
I <- I + 1
Of course, you'll probably want to use for-loops, and += and -= operators, but still the idea is there. For each frame you only need one subtraction, one addition, and one division by a constant (which can be modified into a multiplication or even a bitwise shift in some cases if you are in a hurry).
I may have misunderstood your problem but I think what you're trying to do is the following. Basically, read all images into a matrix and then use mean(). This is providing that you are able to put them all in memory.
function [finish] = stackAndMeanImage (ini_frame, final_frame)
pkg load image;
dir_path = 'C:\Users\Disc-1119\Desktop\Internships\Tracking\Octave\highway\highway (6-13-2014 11-13-41 AM)';
imgs = cell (1, 1, d);
## read all images into a cell array
current_frame = ini_frame;
for n = 1:(final_frame - ini_frame)
fname = fullfile (dir_path, sprintf ("%i", current_frame++));
imgs{n} = rgb2gray (imread (fname, "jpg"));
endfor
## create 3D matrix out of all frames and calculate mean across 3rd dimension
imgs = cell2mat (imgs);
avg = mean (imgs, 3);
## mean returns double precision so we cast it back to uint8 after
## rescaling it to range [0 1]. This assumes that images were all
## originally uint8, but since they are jpgs, that's a safe assumption
avg = im2uint8 (avg ./255);
imwrite (avg, fullfile (dir_path, "meanimage.jpg"));
finish = "done";
endfunction
I am searching for a way to find the location of a button on my screen using matlab. Now i have created the following code for this:
With this code i make a screenshot of my computer, load it in matlab and make it double.
Then do the same with a target image, the image that needs to be found on my screen, and then try to match this to find matrix to the total matrix and display the coordinates of the top left corner of this image.
The problem is that with this method and a screen resolution of 1920*1080 it takes about 15 min to go trough this full proces.
So my question is, is there an easier/faster way to find the coördinates(or the center of) an image on my screen, or within another image? So far i have found none, even when using the image processing toolbox.
clear all
close all
robo = java.awt.Robot;
t = java.awt.Toolkit.getDefaultToolkit();
rectangle = java.awt.Rectangle(t.getScreenSize());
image = robo.createScreenCapture(rectangle);
filehandle = java.io.File('screencapture.png');
javax.imageio.ImageIO.write(image,'png',filehandle);
scrimg=im2double(imread('screencapture.png')); % screenshot
sfimg=im2double(imread('searchfor.png')); % image to look for
[mA,nA,zA] = size(sfimg);
[mB,nB,zB] = size(scrimg);
F = zeros((mB-mA+1)*(nB-nA+1),3);
k = 0;
for p = 1:mB-mA+1
for q = 1:nB-nA+1
for r = 1:zB-zA+1
iets=[p q r];
disp(iets)
if all(all(sfimg==scrimg(p:p+mA-1,q:q+nA-1,r:r+zA-1)))
k = k + 1;
F(k,:) = [p,q,r];
end
end
end
end
F(k+1:end,:) = [];
Check out the following links:
http://repository.asu.edu/items/15110
http://repository.asu.edu/attachments/94025/content//tmp/package-fN0ldX/Velocimeter.pdf
http://repository.asu.edu/attachments/94026/content//tmp/package-fN0ldX/ThesisPresentationr2.pdf
I did 2d tracking from videos in matlab against some very particular geometries. It was radically faster than 15 minutes. The part you are interested in is "measure on image".
The basic idea is that you can use fourier-domain 2d operations, a "max" and some local polynomial fitting to get reliable sub-pixel alignment.
Best of luck
A quicker way is to instead of searching for the entire image is to take a smaller subset of pixels say:
n=4;
SegmentOfImage=sfimg(1:(mA/n),1:(nA/n));
and search for that segment. Obviously as n increases the algorithm would be quicker.
If you want to then make sure that it was actually that segment you can make another comparison on the whole image but now you know the index to start with.
Note:
1) This algorithm can actually take longer if there are a lot of objects with identical pixels to
SegmentOfImage.
2) Also, I believe that this code in general would work only if the matrices are exactly the same. For example if the size of sfimg is not the same as scrimg it would fail.
Hope this helps
i need to clean this picture delete the writing "clean me" and make it bright.
as a part of my homework in image processing course i may use matlab functions ginput, to find specific points in the image (of course in the script you should hard code the coordinates you need).
You may use conv2, fft2, ifft2, fftshift etc.
You may also use median, mean, max, min, sort, etc.
my basic idea was to use the white and black values from the middle of the picture and insert them into the other parts of the black and white strips. however gives a very synthetic look to the picture.
can you please give me a direction what to do ? a median filter will not give good results.
The general technique to do such thing is called Inpainting. But in order to do it, you need a mask of the regions that you want to in paint. So, let us suppose that we managed to get a good mask and inpainted the original image considering a morphological dilation of this mask:
To get that mask, we don't need anything much fancy. Start with a binarization of the difference between the original image and the result of a median filtering of it:
You can remove isolated pixels; join the pixels representing the stars of your flag by a combination of dilation in horizontal followed by another dilation with a small square; remove this just created largest component; and then perform a geodesic dilation with the result so far against the initial mask. This gives the good mask above.
Now to inpaint there are many algorithms, but one of the simplest ones I've found is described at Fast Digital Image Inpainting, which should be easy enough to implement. I didn't use it, but you could and verify which results you can obtain.
EDIT: I missed that you also wanted to brighten the image.
An easy way to brighten an image, without making the brighter areas even brighter, is by applying a gamma factor < 1. Being more specific to your image, you could first apply a relatively large lowpass filter, negate it, multiply the original image by it, and then apply the gamma factor. In this second case, the final image will likely be darker than the first one, so you multiply it by a simple scalar value. Here are the results for these two cases (left one is simply a gamma 0.6):
If you really want to brighten the image, then you can apply a bilateral filter and binarize it:
I see two options for removing "clean me". Both rely on the horizontal similarity.
1) Use a long 1D low-pass filter in the horizontal direction only.
2) Use a 1D median filter maybe 10 pixels long
For both solutions you of course have to exlude the stars-part.
When it comes to brightness you could try a histogram equalization. However that won't fix the unevenness of the brightness. Maybe a high-pass before equalization can fix that.
Regards
The simplest way to remove the text is, like KlausCPH said, to use a long 1-d median filter in the region with the stripes. In order to not corrupt the stars, you would need to keep a backup of this part and replace it after the median filter has run. To do this, you could use ginput to mark the lower right side of the star part:
% Mark lower right corner of star-region
figure();imagesc(Im);colormap(gray)
[xCorner,yCorner] = ginput(1);
close
xCorner = round(xCorner); yCorner = round(yCorner);
% Save star region
starBackup = Im(1:yCorner,1:xCorner);
% Clean up stripes
Im = medfilt2(Im,[1,50]);
% Replace star region
Im(1:yCorner,1:xCorner) = starBackup;
This produces
To fix the exposure problem (the middle part being brighter than the corners), you could fit a 2-D Gaussian model to your image and do a normalization. If you want to do this, I suggest looking into fit, although this can be a bit technical if you have not been working with model fitting before.
My found 2-D gaussian looks something like this:
Putting these two things together, gives:
I used gausswin() function to make a gaus. mask:
Pic_usa_g = abs(1 - gausswin( size(Pic_usa,2) ));
Pic_usa_g = Pic_usa_g + 0.6;
Pic_usa_g = Pic_usa_g .* 2;
Pic_usa_g = Pic_usa_g';
C = repmat(Pic_usa_g, size(Pic_usa,1),1);
and after multiply the image with the mask you get the fixed image.
Hello,
I have a segmented image as shown. Is there a way to smoothen the lines so that it does not look so wavy? Thanks.
The following code requires Image Processing Toolbox:
url = 'http://i182.photobucket.com/albums/x11/veronicafmy/FYP/picture5segmentedimage.jpg';
rgb = imread(url);
bw = im2bw(rgb2gray(rgb), 0.5);
se = strel('line',50,74); % 74 degrees determined by inspection
bw2 = imclose(bw,se);
se2 = strel('line',50,74+90);
bw3 = imclose(bw2,se2);
Here's the result:
Optional step: postprocess by thinning:
bw4 = bwmorph(bw3,'thin',inf);
I think you should ask yourself why it has to be smoother. If you have segmented an image and gotten that result, are you sure that smoothening will give you a correct result?
If it does then Steve Eddins answer seems to do the trick.
If, on the other hand, the object you are trying to segment is much smoother than the result I'd suggest one of two approaches.
If the target object is a cross (two lines), I'd probably calculate the lines and change the representation to two line segments. These can then be rendered at whatever precision and smoothness. To do this you could either find the center and rotation using some kind of feature detection algorithm, or you could use hough transforms to find the lines. The latter is probably much simpler.
If the target can have any form then I'd look into a better segmentation algorithm. There are segmentation algorithms that is not based on hard thresholds. I have used graph partitioning algorithms for this, and while slow, they work well.