Surface registration by markers guided nonrigid iterative closest points algorithm - Emerging Trends in Image Processing, Computer Vision, and Pattern Recognition - page 415

FIGURE 2 Distance map [mm] (left column) and correspondence map [number of units] (right
column) for different modifications of ICP computing correspondence: Euclidean distance (E),
normal shooting with initial rigid registration (NH), static marker vectors (SM), and dynamic
marker vectors (DM).
Because it is difficult to measure directly the quality of correspondences, observation was
proposed in a few steps. Correspondence map ( Figure 2 ) showed spatial distribution of the
feature, number of correspondences assigned to every target point (desirable value is 1). It
is easier to compare correspondence map globally with different cases using correspondence
map histogram ( Figure 3 ). Average correspondence assignment error of points nearest to the
markers allows to measure the quality of correspondence points from cloud, which are nearest
to the markers.
FIGURE 3 Distance map histogram [mm] (a) and correspondence map histogram [number of
units], (b) in different modifications of ICP computing correspondence: Euclidean distance (E),
normal shooting with initial rigid registration (NH), static marker vectors (SM), and dynamic
marker vectors (DM).
The results show improvement when markers are used not only in computing transform-
ation phase but also in finding correspondence phase in every algorithm's iteration [ 9 ] . To
make the proposed changes more universal k-nearest neighbor method and radius constraint
could be used to apply the marker information to not only every point in the cloud but also
the nearest points to the markers. For points that are not near a marker, the Euclidean dis-
tance was used. The score results for three selected values (i.e., 5, 10, and 15%) of the radius
constraint selected as percent of the cloud width were calculated. The results are presented in
Figure 4 . For the most demanding criterion (5%), the results showed improvement.