I have large logical vector looking as V = [0 0 0 0 0 0 0 1 1 1 0 0 1 1 1 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 …………..]
I need to find the position of each group of 1 (lets say – center of each group) but if two groups of ones are too close to each other (say, less than 3 zerros in between) I need to consider those groups as a single group. I.e. at the firs stage I need to find groups (bold-underlined elements) and then find the ceter element of each group (shift +/-1 element does not matter)
1st stage (clusterization): [0 0 0 0 0 0 0 1 1 1 0 0 1 1 1 0 0 0 0 0 0 0 1 1 1 1 1 1 1 0 0 0 0 …………..]
2nd stage (find a center of each cluster): [0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 …………..]
The way I implemented now is following: I do smoothing of the entire vector (it is couple million elements). The span is chousen to be equal of maximum expected lenght of the group and then I look for local maxima (islocalmax) with 'MinSeparation' of minimum distace between groups. It works, but really slow (I have 360×180 = 64800 of vectors – yes, it is LAT/LONG grid with ~10M elements in each vector)
Is any way to speed up this? I believe it should be some "textbook" examples of it!
Best Answer