HOW TO APPLY A MEAN FILTER FOR 3x3

86 次查看（过去 30 天）
ajith
ajith 2012-9-27
评论： DGM 2023-2-10
I NEED TO APPLY THE 3x3 MEAN FILTER TO THE GRAY SCALE IMAGE. PLS SEND ME THE CODE FOR THAT

请先登录，再回答此问题。

采纳的回答

Wayne King
Wayne King 2012-9-27
How about just:
h = 1/3*ones(3,1);
H = h*h';
% im be your image
imfilt = filter2(H,im);
  6 个评论
显示 4更早的评论
Walter Roberson
Walter Roberson 2023-2-10
Check class() of the input image and class() of the output image. I suspect that class of the output is double and that you would get what you want if you uint8() that.
DGM
DGM 2023-2-10
Ran in:
What is the class and scale of the input and output? The rendering of an image depends on an assumption of where white and black are. Floating-point images are expected to be unit-scale (i.e. 0-1). If you're using the above code, then it will be uint8-scale (0-255)
% this is a uint8 image
im = imread('cameraman.tif');
% create filter
h = 1/3*ones(3,1);
H = h*h';
% the output of this will be class 'double', but scaled as if uint8
imfilt = filter2(H,im);
% cast it to a class appropriate for its scale
imfilt = uint8(imfilt);
% display it
imshow(imfilt)
See also:
https://www.mathworks.com/help/images/convert-image-data-between-classes.html

请先登录，再进行评论。

更多回答（5 个）

Image Analyst
Image Analyst 2012-9-27
filteredImage = conv2(single(grayscaleImage), ones(3)/9);
  7 个评论
显示 5更早的评论
Ridho Liwardana
Ridho Liwardana 2022-3-23
It's depend of matrix that you wanna use as filter, if it's a 3x3 matrix, the value is 1/(3x3) -> 1/9.
Image Analyst
Image Analyst 2022-3-23
@SP The reason you divide by windowWidth^2 (1/9 in this case) is because if you don't, the filtering is the sum of the filter window times the image to be filtered. So if the image were all 255 under the filter, you'd get an output value of 9*255 instead of 255. So the image would be 9 times as bright. To ensure that the filtered image lies in the same intensity range as the original image, you need to divide the filter values by windowWidth^2.
If you didn't have a uniform filter (like here where they are all 1's) you'd have to divide by the sum of the kernel:
% Divide unnormalized kernel by the sum of the values in it.
% This will ensure the output image has the same intensity range as the original.
kernel = kernel / sum(kernel(:));
outputImage = imfilter(grayImage, kernel);
Actually, that is the most general approach which will work for any filter.

请先登录，再进行评论。

Mohamed Batran
Mohamed Batran 2015-7-11
thank you for your reply simple but helped a lot for customizing the filter i want
Celal Can Elgün
Celal Can Elgün 2015-10-24
How can we make it without using filter2(or any command) ?
Geetha raja
Geetha raja 2018-8-19
I NEED TO APPLY THE bilateral FILTER TO THE GRAY SCALE IMAGE for denoising. PLS SEND ME THE CODE FOR THAT
  1 个评论
Image Analyst
Image Analyst 2018-8-19
Geetha, there is a part of the web site called "File Exchange". You can look there for community-contributed programs: Search for bilateral filter

请先登录，再进行评论。

Jahid Hasan
Jahid Hasan 2022-4-19
Sorry to ask here, how to write mean and median filter without using a in-built function? What are the way to do it? Any help or code syntax. Thank you
  4 个评论
显示 2更早的评论
Walter Roberson
Walter Roberson 2022-4-19
Potential reference implementation. Not tested.
sum_of_elements = 0;
index_of_element = 0;
while true
index_of_element = index_of_element + 1;
try
sum_of_elements = sum_of_elements + YourVector(index_of_element);
catch ME
break; %we ran off the end of the array
end
end
number_of_elements = index_of_element - 1; %loop over-counts by 1
mean_of_YourVector = sum_of_elements ./ number_of_elements;
sorted_vector = bogosort(YourVector, number_of_elements);
index_of_element = 0;
while true
index_of_element = index_of_element + 1;
if index_of_element + index_of_element == number_of_elements
median_of_YourVector = YourVector(index_of_element);
break
elseif index_of_element + index_of_element + 1 == number_of_elements
median_of_YourVector = (YourVector(index_of_element) + YourVector(index_of_element+1)) / 2;
break
end
end
function sortedVector = bogosort(vectorToSort, number_of_elements)
%this is a legitimate sort
while true
neworder = randperm(number_of_elements);
sortedVector = vectorToSort(neworder);
index_of_element = 0;
while true
index_of_element = index_of_element + 1;
if index_of_element >= number_of_elements
%we got to the end of the list without detecting out-of-order
return;
end
if YourVector(index_of_element) > YourVector(index_of_element+1)
%detected something out of order, try again with a new order
break
end
end
end
end

请先登录，再进行评论。

类别

在 Help Center 和 File Exchange 中查找有关 Image Filtering 的更多信息

标签

产品

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by