D. Contour Extraction
In order to reduce the complexity of the visual
environment to manageable levels, perceptual organization is
used to select parts of the image, or subsets of features that
are of particular interest for further processing. Contour
extraction is one way in which regions of the image that is
likely to correspond to objects of interest can be identified.
We adopt emptying the internal pixel method to extract
contour of the binary image obtained by threshold
segmentation, the principle is assuming the background color
is black, the target color is white, if the original image has a
pixel is white, in addition, whose eight neighboring points
are white, it can be identified as the internal point, which
should be removed [10], the other points are boundary points
make up the contour of object.
Based on this extraction method, the edge of one spur
gear detected as shown in Fig. 2 (c).