Research on an Extracting Method to Assist Manual Identification of Targets - Advances in Image and Graphics Technologies - page 115

3.4 Dilation

b is the structural element, that is the template of dilation. f is the set of images.
The formula of dilation is as the following:

{

[ ]

}

ˆ

f ↕ b = x ( b )

x 

f ≠ ˆ

( 10 )

In the expression, b ˆ is the mapping of b . b ˆ and b are symmetric about origin.

4 Experiment Process and Analysis

4.1 Image Pretreatment

Fig. 1 and Fig. 2 are the results of two processes, which are respectively resolution
reduction and image segmentation. The advantage of the image segmentation method
is retaining all the pixels, which contributes to high processing precision, and the
processing speed is also improved. But when using this method, the rough size of the
target should be known in advance. Besides, the problem that is taking up too much
processing memory is not solved. Although the processing speed is improved, parallel
processing increases the cost of execution. For the larger target, resolution reduction
is better. Since it increases the processing speed, it preserves well the target's contour.
The proposed approach, in this essay, aims to improve the speed of target extraction.
Hence the resolution reduction method is adopted and its result is used for further
processing.
After the resolution reduction, the overall image becomes gloomy and image's de-
tails become blurring. In order to facilitate further processing, the improving of the
visual effect of image is needed.
Fig. 3 and Fig. 4 are the results which are treated respectively with the linear gray
level transformation as well as the histogram equalization method. After being
processed by the means of the histogram equalization, the contrast of the image is
significantly enhanced, but the top right corner is over-bright and blurring. Compari-
son shows that the processing result of the linear gray level transformation is better
because the visual effect of image is effectively improved. Besides, the computational
complexity of the linear gray level transformation is smaller and the processing speed
is faster. In order to achieve the goal, the objects can be extracted quickly and exactly
because the linear gray level transformation method had been adopted.

4.2 Suspected Target Extraction

The binarization result of the double window binarization method depends on the size
of the two windows. According to the experiment, choosing 25 and 50 as the size of
two windows respectively can obtain a relatively better effect and the processing
speed is faster. Fig. 5 is the result image of the double window binarization method.