Toolbox ia870 | List of Figures | Fig. 6.21 | Fig. 6.26

Figure 6.24 - Regional maxima

Description

Typically, a real image presents a large number of regional maxima because of the inherent noise associated with the acquisition process. If the regional maximum operator is applied to a gradient image, then the situation is even worse. Filtering the domes of the image removes regional maxima. Dome filtering can be accomplished using opening, reconstructive opening, area open, or -maxima. The choice of which filter to use is part of the design strategy.

Figure 6.24 shows the regional maxima of the input image following different filters: (a) input image; (b) regional maxima without filtering; (c) regional maxima following opening by a disk; (d) regional maxima following reconstructive opening by a disk; (e) regional maxima following area open; and (f) regional maxima following -maximum. Note that the regional maxima are a binary image and as such the result can be seen as a segmentation strategy. This is an alternative to the common segmentation by thresholding.

Demo Script

 1 import ia870 as MT
 2 
 3 f = adreadgray('MVBook/pinha_n.png')
 4 
 5 B = MT.iasedisk(4)
 6 B8 = MT.iasebox()
 7 
 8 f1 = MT.iaregmax(f,B8)
 9 f2 = MT.iaregmax(MT.iaopen(f,B),B8)
10 f3 = MT.iaregmax(MT.iaopenrec(f,B),B8)
11 f4 = MT.iaregmax(MT.iaareaopen(f,500),B8)
12 f5 = MT.iaregmax(MT.iahmax(f,80),B8)
13 
14 adshow(MT.iapad(f), '(a)')
15 adshow(MT.iapad(MT.ianeg(f1)), '(b)')
16 adshow(MT.iapad(MT.ianeg(f2)), '(c)')
17 adshow(MT.iapad(MT.ianeg(f3)), '(d)')
18 adshow(MT.iapad(MT.ianeg(f4)), '(e)')
19 adshow(MT.iapad(MT.ianeg(f5)), '(f)')

(a)

(b)

(c)

(d)

(e)

(f)