Multiple Plane Detection Method from Range Data of Digital Imaging System for Moving Robot Applications - Visual Computing: Scientific Visualization and Imaging Systems - page 199

Table 11.1 Some angles for data sampling

1st 2nd 3rd 4th 5th 6th 7th 8th 9th 10th

180 205 155 230 130 255 105 280 80 305

Fig. 11.7 The method of
plane segment using IRHT

Because many samplings are required for accurate plane detection, the rotation
of three vertices must be repeatedly calculated. Also, we may fail to get the 3D
coordinates from the measurement error of the sensor system after a rotation for
the original three points. In this case, we try to get the 3D coordinates from another
angle. Table 11.1 shows sequential attempts of a few rotation angles.
Therefore, we can previously save the calculated result of trigonometric
function into the LUT (Look up table) to avoid repeated calculation of the func-
tions. Although additional memory is necessary for building the LUT, it is
indispensable for real-time algorithmic processing.

11.3.4 Plane Segmentation

IRHT (Iterative Randomized Hough Transform) [ 11 ] is introduced for plane
segmentation. Figure 11.7 presents the plane segmentation steps. IRHT is a seg-
mentation method using sequential elimination of maximum peak data from whole
image to reduce the influence of noises and other candidate shapes.
In step , the 3D range data is obtained from the stereo camera. The q, h, and u
values of the sampled data are voted into Hough parameter space, as in step `.