Remove unwanted region in image by matlab -

i have image includes object , unwanted region (small dots). want remove it. hence, use morphological operator example 'close' remove. not perfect. have other way remove more clear? can download example image @ raw image

this code

load image.mat %load img value img= bwmorph(img,'close'); imshow(img); 

you might prefer faster , vectorized approach using bsxfun along information obtained bwlabel itself.

note: bsxfun memory intensive, that's precisely makes faster. therefore, watch out size of b1 in code below. method slower once reaches memory constraints set system, until provides speedup on regionprops method.

code

[l,num] = bwlabel( img ); counts = sum(bsxfun(@eq,l(:),1:num)); b1 = bsxfun(@eq,l,permute(find(counts>threshold),[1 3 2])); newimg = sum(b1,3)>0; 

edit 1: few benchmarks comparisons between bsxfun , regionprops approaches discussed next.

case 1

benchmark code

img = imread('coins.png');%%// 1 chosen available in matlab image library img = im2bw(img,0.4); %%// 0.4 seemed make enough blobs image  lb = bwlabel( img ); threshold = 2000;  disp('--- regionprops method:'); tic,out1 = regionprops_method1(img,lb,threshold);toc clear out1  disp('---- bsxfun method:'); tic,out2 = bsxfun_method1(img,lb,threshold);toc  %%// demo, have rejected enough unwanted blobs figure, subplot(211),imshow(img); subplot(212),imshow(out2); 

output

benchmark results

--- regionprops method: elapsed time 0.108301 seconds. ---- bsxfun method: elapsed time 0.006021 seconds. 

case 2

benchmark code (only changes case 1 listed)

img = imread('snowflakes.png');%%// 1 chosen available in matlab image library img = im2bw(img,0.2); %%// 0.2 seemed make enough blobs image threshold = 20; 

output

benchmark results

--- regionprops method: elapsed time 0.116706 seconds. ---- bsxfun method: elapsed time 0.012406 seconds. 

as pointed out earlier, have tested other bigger images , lot of unwanted blobs, bsxfun method doesn't provide improvement on regionprops method. due unavailability of such bigger images in matlab library, couldn't discussed here. sum up, suggested use either of these 2 approaches based on input features. interesting see how these 2 approaches perform input images.