Research Article

A Set of Geometric Features for Neural Network-Based Textile Defect Classification

Figure 10

Important part of 5 1 2 × 5 1 2 -pixel image of all types of defect. (a) For color yarn, T A D R = 9 3 5 p i x e l s . (b) For hole, T A D R = 8 7 4 p i x e l s . (c) For hole, T A D R = 4 2 6 9 p i x e l s . (d) For vertical missing yarn, T A D R = 4 6 1 1 p i x e l s . (e) For vertical missing yarn, T A D R = 1 0 5 1 3 p i x e l s . (f) For horizontal missing yarn, T A D R = 4 4 8 3 p i x e l s . (g) For horizontal missing yarn, T A D R = 1 0 4 8 6 p i x e l s . (h) For spot, T A D R = 4 1 5 4 p i x e l s . (i) For spot, T A D R = 1 2 8 0 0 p i x e l s .
643473.fig.0010a
(a)
643473.fig.0010b
(b)
643473.fig.0010c
(c)
643473.fig.0010d
(d)
643473.fig.0010e
(e)
643473.fig.0010f
(f)
643473.fig.0010g
(g)
643473.fig.0010h
(h)
643473.fig.0010i
(i)