Efficient BFCN for Automatic Retinal Vessel Segmentation

Figure 2

Preprocessing effect diagram. The first row is the DRIVE dataset, the second row is the STARE dataset, and the third row is the CHASE dataset. (a) The original image of each dataset, (b) the corresponding real ground truth, (c) the red (R) channel image of the original image, (d) the green (G) channel image of the original image, (e) the original image for the blue (B) channel image, (f) a preprocessed three-channel image, and (g) a process for converting the preprocessed three-channel image into a grayscale image.

(a)

(b)

(c)

(d)

(e)

(f)

(g)