1302
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B4. Beijing 2008
. .
taining characters stable of the whole image and big bright area.
Whereas Morphological close operation could remove dark de
tails which is smaller than structure element and maintain char
acters stable of the whole image and big dark area. Structure
element with circle shape has good suitability. Taking all above
factors we design our morphological filter algorithm like fol
lows:
Step 1: Employ circular structural element which is bigger
then the target, training it with background samples, after the
Morphological opening operation noise and target both will be
removed. The final result is considered as the background.
Step 2: The opening operation is acted on the primary image
with the structural element which was trained by target samples.
The bright noise whose size is smaller then the structure ele
ment is filtrated .And the background is consistent with Step 1.
Thanks to the trained structure elements own target information,
target will not be filtrated
Step 3: The result of the image from the step 2 subtracts the
background images in step 1 is the final target.
4. MORPHOLOGY ALGORITHM OPTIMIZED BY
GENETIC THEORY
The collectivity flow chart of target detection algorithm is
shown in Figure 1. Figure 2 is a detailed introduction to the
“Genetic Evolution” part of Figure 1
Figure 2 Genetic evolution algorithm flow chart
Figure 1 Optimized morphology target detection algorithm
by genetic theory
5. THE APPLICATION OF ALGORITHM
In order to verify the validity of the algorithm, in this paper,
two kinds of natural background infrared image were used to
the experiment.
a. single objective infrared image b. multi-objective infrared
image c. salt and pepper noise image
Figure 3.Original infrared image