Edge Detection from RGB-D Image Based on Structured Forests

Algorithm 1

Growth randomized tree ().

Input: training sample set , ,

Output: random tree classifier

if all the training samples of belong to the same category or , then

  return

end if

select parameter space subset randomly:

for to do

end for

Compute the optimal parameter of the node classifier:

Set the current dataset of the left and right child nodes : ,

for to do

   if then

   else

   end if

end for

New left child node:

New right child node:

return