Image pre-processing to remove noise

Hi. Is there any image pre-propcessing technique to remove noise from binary images? Please see attached image, I want to remove the objects that are not very straight lines and are connected together, a cxongregation of pixels that look like small circles in the image.

Any hints with MATLAB commands can really help.

Thanks.

Answer 1

[KALYAN ACHARJYA]

Absolutely possible, you can remove the white pixels based on area (Extract objects from binary image by size) and do some AND or OR logical Operation to get the resultant image.

https://in.mathworks.com/help/images/ref/bwareafilt.html

[SS]

Hi, thanks for youre reply. I am getting this error:

Error using bwpropfilt

Expected input number 1, BW, to be one of these types:

logical

Instead its type was uint8.

Error in bwpropfilt>parse_inputs (line 121)

validateattributes(bw, {'logical'}, {'nonsparse', '2d'}, mfilename, 'BW', 1);

Error in bwpropfilt (line 58)

[bw, I, attrib, p, direction, conn] = parse_inputs(args{:});

Error in bwareafilt (line 34)

bw2 = bwpropfilt(bw, 'area', p, direction, conn);

[KALYAN ACHARJYA]

data=~bwareafilt(input_binary_image,20);
%........................^ Set as per requirement  
result=input_binary_image & result;
figure,imshow(result);

Or see regionprops function and do accordingly

[Image Analyst]

How did you get your binary image, bw? By thresholding your original gray scale image? It should be logical then. You can't call it on a gray scale image.

bw = grayImage < 100; % or whatever
bw2 = bwpropfilt(bw, 'area', p, direction, conn);

[SS]

The raw binary image is obtained by thresholding and combining few gray images.