Fig.1

Picture of fabric with warp defect.(a) Linear defect; (b) Hole shape defect"

Fig.2

Sample images. (a) Linear defect; (b) Hole shape defect"

Fig.3

Algorithm flow chart"

Fig.4

Boolean difference set operation between original graph and eigenvalue graph. (a) Defects in original image; (b) Defect eigenvalue diagram; (c) Defect difference chart"

Fig.5

Mean filtering and histogram averaging. (a) Defect mean filtering; (b) Histogram averaging"

Fig.6

Variance threshold filter kernel"

Fig.7

Variance threshold filtering.(a) Linear defect; (b) Hole shape defects"

Fig.8

Schematic of expansion and corrosion operations.(a) Erosion operation; (b) Dilation operation"

Fig.9

Morphological processing. (a) Defect opening operation; (b) Defect expansion operation"

Fig.10

Detection results of hole defects. (a) Area threshold filtering; (b) Defect original drawing verification"

Fig.11

Linear defects are compared with the eigenvalue diagrams of normal fabrics. (a) Original image; (b) Eigenvalue image"

Fig.12

Linear defects and normal fabric variance threshold filtering results. (a) Eigenvalue image of fabric; (b) Image of variance threshold filtering"

Tab.1

Comparison of three fabric defect algorithms"

Fig.13

Comparison diagram of hole defects. (a) Original image; (b) Method of reference [1]; (c) Method of reference[8]; (d) Method of this paper; (e) Detection results of this paper"

Fig.14

Linear defect comparison diagram. (a) Original image; (b) Method of reference [1]; (c) Method of reference[8]; (d) Method of this paper; (e) Detection results of this paper"