See this snippet:
img = uint8(randi(256, [800,600])-1); % Synthesize an image.
% Get the histogram.
[counts grayLevels] = imhist(img);
a = 4
b = 9
% Find the minimum count in the range [a,b]
minCountInRange = min(counts(a:b))
% Get a logical vector that says if the count is
% the min count value or not.
indexesWithCount = (counts == minCountInRange)
% Filter to get only those gray levels in the range [a,b]
% Now get the actual index number(s):
indexOfMin = find(indexesWithCount(a:b)) + a - 1
gls = grayLevels(indexOfMin)
% Note: gray level, say, 6 occurs at index 7 because
% a gray level of 0 starts at index 1, not index 0.
