Edge detection model based on multi-expert and principal component analysis

Abstract:

Each method of edge detection is regarded as an expert. To solve the problem of inconsistency of different expert responses and the difficulty of choosing right expert, a model of edge detection based on multi-expert and principal component analysis (PCA) is proposed. Firstly, a statistical model for expert response is established, and theoretical analysis shows that the edge detection model can effectively suppress noise coming from the expert. Then, the information of multiple detection results were combined to obtain the general detection result with the model based on multi-expert and PCA. The experimental results are very satisfactory.