Background:
In the nickel foam production process, the detection and identification of
surface defects relies heavily upon the operators’ experiences. However, the manual observation is
of high labor intensity, low efficiency, strong subjectivity and high error rate.
Objective:
Therefore, this paper proposes a new method for the nickel foam surface defect detection
and identification, based on an improved probability extreme learning machine.
Methods:
At first, a machine vision system for nickel foam is established, and gray level cooccurrence
matrix is used to calculate defect features, which are inputted into extreme learning machine
to train the defect classifier. Then a composite differential evolution algorithm is used to optimize
the input weights and hidden layer thresholds. Finally, an integrated probabilistic ELM is
proposed to avoid misjudgments when multiple probabilities values are almost identical.
Conclusion:
Experiments show that the proposed method can achieve a defect-identifying accuracy,
which meets an enterprise’s needs.