Image segmentation with region-based thresholds in Matlab -

i have noisy image multiple separated circular regions blurred out. example of such image 6 region of interests (roi) :

image source

segmenting image global threshold easy in matlab using bwconncomp , given threshold. want set fix threshold (e.g. 54%) with respect maximum pixel value of every roi (instead of whole image) to segment each roi.

i have collection of images different roi sizes , positions, , need segment them based on regional-based thresholding, therefore cannot use matlab interactive tool select them either.

thanks

after you've done connected component analysis try using bwlabel. label each of 6 regions number 1-6. use each of regions/labels mask. @ values in region , find max

%after bwconncomp lbls = bwlabel(my_bw_conn_output); num_lbls = max(lbls(:));  percentage_of_max = 54; my_thresh = zeros(1,num_lbls);  ii=1:num_lbls     %isolates 1 roi     temp_mask = (lbls == ii);      %we 2 maxes, 1 rows 1 columns     max_val = max(max(uint8(temp_mask) .* image_with_values_to_threshold));      %this color pictures     my_thresh(ii) = percentage_of_max * max_val end 

edit

if want use results of bwconncomp directly more this

%my code written assuming started grayscale image, though %could adapt color image cc = bwconncomp(grayscale_image); percentage_of_max = 54;  ii=1:cc.numobjects     mask = zeros(cc.imagesize);      %converts indexes x,y coordinates     [x_coords,y_coords] = ind2sub(cc.imagesize,cc.pixelidxlist{ii}(:));     mask(x_coords,y_coords) = 1;      %masks image     roi_im = bsxfun(@times, uint8(mask), grayscale_image);      %% processing region here using roi_im      curr_thresh = max(roi_im (:)) * percentage_of_max;      thesh_im = grayscale_image > curr_thresh; end 

another alternative crop images , run calculations on cropped image. jsut came across post how crop face section image given corner points. matlab helps cropping perfect squares. can use roipoly if have shape isn't rectangle.