Image Processing Reference
In-Depth Information
[p2,msk2]=phaseproc(dir2,TH,FILTER);
dp=p1-p2;
msk=msk1.*msk2;
dp=dp.*msk;
6.5 Noise Removal from Differential Phase
As can be seen, the phase map ( Figure 18 ) has very low contrast and object topography is not
clear. This is because of phase mistakes with very large or very small values. Hence, we im-
proved the previous phase map using the following ilter:
dp(abs(dp)>15 | dp>-4)=0;
dpf=abs(dp);
After erasing these mistakes, we can see the much improved object topography in the phase
map of Figure 19 .
FIGURE 19 Phase map after filtering.
There are yet some phase errors especially at the edges. We can correct these errors using a
ilter previously used in phase computation. Like with phase computation, we can apply this
ilter several times by which we obtain the further improved phase map of Figure 20 .
FIGURE 20 Final phase map with modulated height information.
a=medfilt2(dpf,[1 7]);
7]);c=filter2([1 2 3 0 3 2 1]/12,dpf);
b=abs(a-dpf)>.5;
dpf=dpf.*~b+c.*b;
6.6 How to Make RGB Texture Image from Projected Fringe
Pattern Images
RGB=(projected fringe patern1+projected fringe patern2+projected fringe patern3)/3
As can be seen from the texture image of Figure 21 , there are some unwanted periodic pat-
terns and these could be removed by using phase filtering techniques in the frequency do-
main. However, a beter method is possible as a result of binary code patern unwrapping be-
 
 
Search WWH ::




Custom Search