updated code for more flexible parallelisation : TBB parallel for loops are replaced by opencv parallel_for_ wrapper