I have a an image of size 180x220 containing some noise in the region for example (145:180,1:65).
My question is how to remove the noise in this region without affecting the other parts of the image using Matlab.
Thank you very much.
Edit: I want to remove the noise in the regions (1:146,1:25) and (1:15,25,174) from the following image:
In general, this would go something like
% filter image in-place
img(145:180, 1:65) = medfilt2(img(145:180, 1:65));
Note that most filters require some context of the region of interest to do a proper interpolation/averaging/etc., so you might want to take this approach:
% Note: increase ROI by 10 on each side
offset = 10;
img_tmp = img(145-offset : 180+offset, 1 : 65+offset);
% apply filter
img_tmp = medfilt2(img_tmp, [additional parameters]);
% put filtered image back in its proper place
img(145:180, 1:65) = img_tmp(offset:end-offset+1, 1:end-offset+1);
img = double(imread('img.jpg'));
h = fspecial('gaussian', hsize, sigma); % decide how to filter the image
img_filt = imfilter(img, h, 'replicate');
now, use the filtered image only in the noise region
img(145:180,1:65,:) = img_filt(145:180,1:65,:);
Edit: After you posted the image I guess you want simply replace the noised region by the vanilla color?
If so, then do the following (assuming gray image):
med_pixel = median(img(:)); % detect the dominant color
img(1:146,1:25) = med_pixel;
img(1:15,25,174) = med_pixel;
... and so on
Related
I have an image that has multi frequency noise, I used the code in this link :
Find proper notch filter to remove pattern from image
source image : orig_image
But my final image noise has not been removed.
As you know, I must remove the gradient in vertical direction. the frequency representation of image come in below:
fft of image
Have anyone idea for removal of this noise in matlab?
I apply sobel filter and median filter but not improved.
Note that my goal is remove of lines inside of object.
Regards.
You have two kinds of noise: irregular horizontal lines and salt and pepper. Getting rid of the lines is easy as long as the object doesn't cover the whole horizontal range (which it doesn't in your example).
I just sample a small vertical stripe on the left to only get the stripes and then subtract them from the whole image. Removing the salt and pepper noise is simple with a median filter.
Result:
Code:
% read the image
img = imread('http://i.stack.imgur.com/zBEFP.png');
img = double(img(:, :, 1)); % PNG is uint8 RGB
% take mean of columsn 100..200 and subtract from all columns
lines = mean(img(:, 100:200), 2);
img = img - repmat(lines, 1, size(img, 2));
% remove salt and pepper noise
img =medfilt2(img, [3,3], 'symmetric');
% display and save
imagesc(img); axis image; colormap(gray);
imwrite((img - min(img(:))) / (max(img(:)) - min(img(:))), 'car.png');
here is code I once used for an assignment I once had to do. It was a much simpler example than the one you have. There were only a total of 4 components in the frequency domain causing noise, so simply setting those 4 components to zero manually (hfreq) solved my problem. I dont know how well this will work in your case, perhaps writing an algorithm to find the appropriate hfreqs that stand out will help. Here was the original image I used :
This is the code I used :
% filtering out the noisy image
clc
clear all
close all
image = imread('freqnoisy.png');
image = double(image);
image = image/(max(max(image)));
imshow(image) % show original image
Fimage = fft2(image);
figure
imshow(((fftshift(abs(Fimage)))),[0 5000]) %shows all frequency peaks that stick out
hfreq = ones(256);
hfreq(193,193)=0;
hfreq(65,65) = 0;
hfreq(119,105) = 0;
hfreq(139,153)= 0;
%
Fimage_filtered = fftshift(Fimage).*hfreq;
figure
imshow(abs(Fimage_filtered),[0 5000]) % show freq domain without undesired freq
filtered_im = ifft2(ifftshift(Fimage_filtered));
figure
imshow(filtered_im)
this is what your output will look like :
I need to find similar corners in image (for example: 4 corners of a rectangle, same corners, just different orientation?).
I have this code:
% read the image into MATLAB and convert it to grayscale
I = imread('s2.jpg');
Igray = rgb2gray(I);
figure, imshow(I);
% We can see that the image is noisy. We will clean it up with a few
% morphological operations
Ibw = im2bw(Igray,graythresh(Igray));
se = strel('line',3,90);
cleanI = imdilate(~Ibw,se);
figure, imshow(cleanI);
% Perform a Hough Transform on the image
% The Hough Transform identifies lines in an image
[H,theta,rho] = hough(cleanI);
peaks = houghpeaks(H,10);
lines = houghlines(Ibw,theta,rho,peaks);
figure, imshow(cleanI)
% Highlight (by changing color) the lines found by MATLAB
hold on
After running this code I convert my starting image into a binary image with:
binary = im2bw(I);
after this I get a product from those 2 binary images and I think I get corners..
product = binary .* cleanI;
now I imfuse this picture with grayscale starting picture and get this:
I dont know what to do to get only those 4 corners!
Ok second try. Below a code that does not finally do the job but it might help.
Edge identifies the contours and with regionprops you get the characteristica of each identified element. As soon as you know what characteristics your desired object has you can filter it and plot it. I went through the Areas in shapedata.Area and the 6th largest was the one you were searching for. If you combine Area with some of the other charateristica you might get the one you want. As I said not ideal and final but perhaps a start ...
clear all
close all
source = imread('Mobile Phone.jpg');
im = rgb2gray(source);
bw = edge(im ,'canny',[],sqrt(2));
shapedata=regionprops (bwlabel(bw,8),'all');
%index = find([shapedata.Area]== max([shapedata.Area]));
index = 213;
data = shapedata(index).PixelList;
figure
imshow(im)
hold on
plot(data(:,1),data(:,2),'ro');
I have an image shown below:
I am applying some sort of threshold like in the code. I could separate the blue objects like below:
However, now I have a problem separating these blue objects. I applied watershed (I don't know whether I made it right or wrong) but it didn't work out, so I need help to separate these connected objects.
The code I tried to use is shown below:
RGB=imread('testImage.jpg');
RGB = im2double(RGB);
cform = makecform('srgb2lab', 'AdaptedWhitePoint', whitepoint('D65'));
I = applycform(RGB,cform);
channel1Min = 12.099;
channel1Max = 36.044;
channel2Min = -9.048;
channel2Max = 48.547;
channel3Min = -53.996;
channel3Max = 15.471;
BW = (I(:,:,1) >= channel1Min ) & (I(:,:,1) <= channel1Max) & ...
(I(:,:,2) >= channel2Min ) & (I(:,:,2) <= channel2Max) & ...
(I(:,:,3) >= channel3Min ) & (I(:,:,3) <= channel3Max);
maskedRGBImage = RGB;
maskedRGBImage(repmat(~BW,[1 1 3])) = 0;
figure
imshow(maskedRGBImage)
In general, this type of segmentation is a serious research problem. In your case, you could do pretty well using a combination of morphology operations. These see widespread use in microscopy image processing.
First, clean up BW a bit by removing small blobs and filling holes,
BWopen = imopen(BW, strel('disk', 6));
BWclose = imclose(BWopen, strel('disk', 6));
(you may want to tune the structuring elements a bit, "6" is just a radius that seemed to work on your test image.)
Then you can use aggressive erosion to generate some seeds
seeds = imerode(BWclose, strel('disk', 35));
which you can use for watershed, or just assign each point in BW to its closest seed
labels = bwlabel(seeds);
[D, i] = bwdist(seeds);
closestLabels = labels(i);
originalLabels = BWopen .* closestLabels;
imshow(originalLabels, []);
I would try the following steps:
Convert the image to gray and then to a binary mask.
Apply morphological opening (imopen) to clean small noisy objects.
Apply Connected Component Analysis (CCA) using bwlabel. Each connected component contains at least 1 object.
These blue objects really look like stretched/distorted circles, so I would try Hough transform to detect cicles inside each labeled component. There is a built-in function (imfindcircles) or code available online (Hough transform for circles), depending on your Matlab version and available toolboxes.
Then, you need to take some decisions regarding the number of objects, N, inside each component (N>=1). I don't know in advance what the best criteria should be, but you could also apply these simple rules:
[i] An object needs to be of a minimum size.
[ii] Overlaping circles correspond to the same object (or not, depending on the overlap amount).
The circle centroids can then serve as seeds to complete the final object segmentation. Of course, if there is only one circle in each component, you just keep it directly as an object.
I didn't check all steps for validity in Matlab, but I quickly checked 1, 2, and 4 and they seemed to be quite promising. I show the result of circle detection for the most difficult component, in the center of the image:
The code I used to create this image is:
close all;clear all;clc;
addpath 'circle_hough'; % add path to code of [Hough transform for circles] link above
im = imread('im.jpg');
img = rgb2gray(im);
mask = img>30; mask = 255*mask; % create a binary mask
figure;imshow(mask)
% filter the image so that only the central part of 6 blue objects remains (for demo purposes only)
o = zeros(size(mask)); o(170:370, 220:320) = 1;
mask = mask.*o;
figure;imshow(mask);
se = strel('disk',3);
mask = imopen(mask,se); % apply morphological opening
figure;imshow(mask);
% check for circles using Hough transform (see also circle_houghdemo.m in [Hough transform for circles] link above)
radii = 15:5:40; % allowed circle radii
h = circle_hough(mask, radii, 'same', 'normalise');
% choose the 10 biggest circles
peaks = circle_houghpeaks(h, radii, 'nhoodxy', 15, 'nhoodr', 21, 'npeaks', 10);
% show result
figure;imshow(im);
for peak = peaks
[x, y] = circlepoints(peak(3));
hold on;plot(x+peak(1), y+peak(2), 'g-');
end
Some spontaneous thoughts. I assume the ultimate goal is to count the blue corpuscles. If so, I would search for a way to determine the total area of those objects and the average area of a corpuscle. As a first step convert to binary (black and White):
level = graythresh(RGB);
BW = im2bw(RGB,level);
Since morphological operations (open/close) will not preserve the areas, I would work with bwlabel to find the connected components. Looking at the picture I see 12 isolated corpuscles, two connected groups, some truncated corpuscles at the edge and some small noisy fragments. So first remove small objects and those touching edges. Then determine total area (bwarea) and the median size of objects as a proxy for the average area of one corpuscle.
I have an image with very low contrast, from which I would like to extract a text object. Since it has a low contrast, I tried a few methods but no method gave me a satisfying result. I used watershed to extract the text object, but as the contrast is poor the extraction was not successful.
My program for watershed is:
I_cropped=imread(strcat('C:\Id\',currentfilename));
I_cropped = rgb2gray(I_cropped);
I_eq = histeq(I_cropped);
figure,imshow(I_eq);
bw = im2bw(I_eq, graythresh(I_eq));
bw2 = imfill(bw,'holes');
bw3 = imopen(bw2, ones(5,5));
bw4 = bwareaopen(bw3, 40);
bw = im2bw(I_eq, graythresh(I_eq));
figure,imshow(bw);
mask_em = imextendedmax(I_eq, 30);
mask_em = imclose(mask_em, ones(5,5));
mask_em = imfill(mask_em, 'holes');
mask_em = bwareaopen(mask_em, 40);
figure,imshow(mask_em);
I_eq_c = imcomplement(I_eq);
figure,imshow(I_eq_c);
I_mod = imimposemin(I_eq_c, ~bw4 | mask_em);
figure,imshow(I_mod);
L = watershed(I_mod);
figure,imshow(label2rgb(L));
I applied laplacian filter and to enhance edge, but it was not effective.
My objective is to extract text object. What method should I try for such low contrast image?
The image is attached:
Here is a way to do it.
First apply a median filter with a large kernel on the image to remove outliers and then apply a threshold to convert to a binary image. Note that playing with the kernel size of the filter alters the threshold level you need to use. Play around with it to see the output changing.
Then invert the image and apply regionprops to detect objects in the image. After that a little math to deduce the x and y origin (defining upper left corner) as well as the width and length of the large bounding box enclosing all the letters from the image.
Here is the code:
clear
clc
close all
Im = rgb2gray(imread('Imtext.png'));
%// Apply median filter to remove outliers
Im = medfilt2(Im,[9 9]);
%// Clear borders and apply threshold, then invert image
ImBW = imclearborder(~(im2bw(Im,.55)));
%// Find bounding boxes to delineate text regions.
S = regionprops(ImBW,'BoundingBox');
%// Concatenate all bounding boxes and obtain x,y, width and length of big
%// bounding box enclosing everything
bb = cat(1,S.BoundingBox);
LargeBB_x = min(bb(:,1));
LargeBB_y = min(bb(:,2));
LargeBB_height = max(bb(:,4));
%// Find last column in which pixel value is 1; that's the end
%// of the bounding box.
[~,ic] = find(ImBW==1);
MaxCol = max(ic(:));
LargeBB_width = MaxCol-LargeBB_x;
%// Display boxes
imshow(ImBW)
hold on
rectangle('Position',[LargeBB_x LargeBB_y LargeBB_width LargeBB_height],'EdgeColor','r','LineWidth',2)
hold off
And the output:
Or with the original image:
I want to use mathematical morphology function in MATLAB to find the boundary of can.png image. The input image is:
I want to get a boundary such as :
I tried to use different combination and parameter using strel, imerode, imdilate , But the result is not good enough (far from the expectation)
One of my trial code is:
a = imread ('can.png');
b = im2bw(a);
SE = strel('rectangle', [10 50 ]) ;
i2 = imdilate(b,SE);
figure(1); imshow(i2);
p = ones(4);
c = b - imerode(b,p);
figure(2); imshow(c);
The output is:
Can any body help me, how to create the expected image (black background with thin boundary for the can, please? Thank you very much.
Binarize on its morphological gradient, then do a dilation with an elementary SE, fill holes and finally obtain its border (trivial given the current image). This doesn't require any magical arbitrary threshold.
im = imread('can.png');
% turn image to BW
imb = (im < 220);
% fill the holes in the image
imf = imfill(imb, 'holes');
% find the edge
ed = edge(imf);
Resulting image: