In this paper, a highly effective and efficient ensemble learning based parallel impulse noise detection algorithm is
proposed. The contribution of this paper is three-fold. First, we propose a novel intensity homogeneity metric-
Directional Homogeneity Descriptor(DHD), which has very powerful discriminative ability and has been proven
in our previous work. Second, this proposed algorithm has high parallelism in feature extraction stage, classifier
training, and testing stage. And the proposed architecture are very suitable for distributed processing. Finally,
instead of manually tune the thresholds for each feature as most of the works in this research area do, we
use Random Forest to make decision since it has been demonstrated to own better generalization ability and
performance comparable to SVM or Boosting in classification problem. Another important reason we adopt
Random Forest is that it has natural parallelism structure and very significant performance advantage (e.g. the
overhead of training and testing the model is very low ) over other popular classifiers e.g. SVM or Boosting.
To the best of our knowledge, this is the first time ensemble learning strategies have been used in the area
of switching median filtering. Extensive simulations are carried out on several most common standard testing
images. The experimental results show that our algorithm achieves zero miss detection results while keeping the
false alarm rate at a rather low level and has great superiority over other state-of-the-art methods.