Oil spill feature selection and classification using decision tree forest on SAR image data

摘要: Abstract A novel oil spill feature selection and classification technique is presented, based on a forest of decision trees. The parameters the two-class problem spills look-alikes are explored. contribution to final 25 most commonly used features in scientific community was examined. work sought framework multi-objective problem, i.e. minimization input and, at same time, maximization overall testing accuracy. Results showed that optimum contains 70 trees three important combinations contain 4, 6 9 features. latter combination can be seen as appropriate solution study. Examination robustness above result proposed achieved higher accuracy than other well-known statistical separation indexes. Moreover, comparisons with previous findings converge (up 84.5%) number selected features, but diverge actual This observation leads conclusion there not single combination; several sets exist which least some critical