Industrial Printing Image Defect Detection Using Multi-Edge Feature Fusion Algorithm

Figure 6

The similarity matching results of different pattern types. (a) The qualified pattern matches the qualified patterns. (b) The qualified pattern matches the defective patterns. (c) The defective pattern matches the qualified patterns. (d) The defective pattern matches the defective patterns. The horizontal axis represents the number of similarity matches. The vertical axis represents the similarity matching results (similarity matching score). In (a), there are 300 red points, and each point represents a matching result. The block pattern without superposition is a qualified pattern. The section superimposed by three block patterns represents the other 300 qualified patterns. In (b), there are 135 red points, and each point represents a matching result. The block pattern without superposition is a qualified pattern. The section superimposed by three block patterns represents the other 135 defective patterns. In (c), there are 300 red points, and each point represents a matching result. The block pattern without superposition is a defective pattern. The section superimposed by three block patterns represents the other 300 qualified patterns. In (d), there are 135 red points, and each point represents a matching result. The block pattern without superposition is a defective pattern. The section superimposed by three block patterns represents the other 135 defective patterns.

(a)

(b)

(c)

(d)