Anagram Engineering

Webdesign und Softwarelösungen aus Vorarlberg. Ihr Partner für innovative Lösungen rund ums Internet. Full Service Agentur.

Line Detection



Next: Synthetic data
Up: Results
Previous: Calibration

Line Detection

The accuracy of the line detection is measured with synthetic and real data and defined by the difference of the manually and the automatic detection of line properties. The properties that are evaluated are

line origin
The coordinate of the first point
line end
The coordinate of the last point
attitude
The attitude of the line

The attitude $ alpha$ is the most important property when it comes to correspondence analysis and is defined as

begin{equation*}begin{aligned}alpha = frac{x_l}{y_l} end{aligned}end{equation*}


where
$ x_{l} = l_{x_e} - l_{x_s}$ and
$ y_{l} = l_{y_e} - l_{y_s}$. $ l_{C_T}$ are the coordinates of the line, $ C$ defines if $ x$ or $ y$ is used and $ T$ defines if it is the start or end point. Note that $ alpha$ is not defined when $ y_l$ is 0. In this special case $ alpha$ is set to $ x_l$. But $ y_l$ is only 0 in the case of a horizontal lines, which cannot be used for the further computation. Due to the importance of the attitude in the following correspondence analysis the images have to be rectified or be the result of a parallel camera system4.2 in order to solve the corresponding problem correctly.

Subsections

Next: Synthetic data
Up: Results