An Improved Edge Detection Method Based on Composite Morphology - Multimedia and Signal Processing - page 115

Therefore, we can use two or three of the above methods according to the specific
image to construct a more comprehensive composite morphology. The method of
edge extraction in this paper based on multi-faceted morphology and the multi-
structure morphology, the algorithm is as follows:
(1)Select M different shape of the structural elements to construct multi-structure
elements set{B M } and select N different directions of linear structural elements to
construct multi-faceted structural elements set {B N };
(2) To the structural elements in the two structural elements set, we use the anti-
noise dilation and erosion morphological gradient operator to extract edge.
(3) We synthesize new edge by the weights of obtained edge image:

K


=

G = w

n g

(3)

n

n 1

G is the synthetic image edge, K is the number of extracted edge (K = M + N), g n is
the extracted edge from the respective structural element, w n is the weights for each
edge.
Figure 3 shows the results of extracted edge by the multi-faceted and multi-
structure morphology.

(a) (b) (c) (d)

Fig. 3. Composite morphological edge extraction (a) Input image (b)Results of multi-structure
morphology (c)Results of the multi-faceted morphology (d) Results of the multi-faceted and
multi-structure morphology

For the above-mentioned composite morphological operations, we are using the
anti-noise dilation and erosion morphological gradient operator to extract edge as
show in formula (2). The operator is actually obtained by combining the
morphological filtering and morphological gradient edge extraction, so it has strong
anti-noise ability and good edge extraction results.
Actually ,the gradient operator used in the two operations is a single structural
elements .As we known, the advantages of multi-structure morphology is able to
extract the complex image edge and have strong noise suppression capability , the
advantages of multi-faceted morphology is good at retaining the edge direction
information and accessing to get the more delicate and smooth edges.
Therefore, we can improve multi-faceted and multi-structure morphology
algorithm. That is the multi-structure morphology play a major role in noise