Line features
These methods extract features such as histograms and moments from grey-scale or color images.
- class machinevisiontoolbox.ImageLineFeatures.ImageLineFeaturesMixin[source]
Line features are common in in many human-built environments.
Supporting classes
- class machinevisiontoolbox.ImageLineFeatures.Hough(image, ntheta=180, drho=1)[source]
Hough line features
- Parameters
image (
Image) – greyscale imagentheta (int, optional) – number of steps in the
thetadirection, defaults to 180drho (int, optional) – increment size in the
rhodirection, defaults to 1
Create a Hough line feature object. It can be used to detect:
lines using the classical Hough algorithm
linesline segments using the probabilistic Hough algorith
lines_p
The Hough accumulator is a 2D array that counts votes for lines
\[u \cos \theta + v \sin \theta = \rho\]with quantized parameters \(\theta\) and \(\rho\). The parameter \(\theta\) is quantized into
nthetasteps spanning the interval \([-\pi, \pi)\), while \(\rho\) is quantized into steps ofdrhospanning the vertical dimension of the image.Note
Lines are not detected until
linesorlines_pis called. This instance simply holds parameters.- Reference
Robotics, Vision & Control for Python, Section 12.2, P. Corke, Springer 2023.
- Seealso
- lines(minvotes)[source]
Get Hough lines
- Parameters
minvotes (int) – only return lines with at least this many votes
- Returns
Hough lines, one per row as \((\theta, \rho)\)
- Return type
ndarray(n,2)
Return a set of lines that have at least
minvotesof support. Each line is described by \((\theta, \rho)\) such that\[u \cos \theta + v \sin \theta = \rho\]- Seealso
- lines_p(minvotes, minlinelength=30, maxlinegap=10, seed=None)[source]
Get probabilistic Hough lines
- Parameters
minvotes (int) – only return lines with at least this many votes
minlinelength (int) – minimum line length. Line segments shorter than that are rejected.
maxlinegap (int) – maximum allowed gap between points on the same line to link them.
- Returns
Hough lines, one per row as \((u_1, v_1, u_2, v_2)\)
- Return type
ndarray(n,4)
Return a set of line segments that have at least
minvotesof support. Each line segment is described by its end points \((u_1, v_1)\) and \((u_2, v_2)\).- Seealso
plot_lines_plines
- plot_lines(lines, *args, **kwargs)[source]
Plot Hough lines
- Parameters
Detected lines are given as rows of
lines:for Hough lines, each row is \((\theta, \rho)\), and lines are clipped by the bounds of the current plot.
for probabilistic Hough lines, each row is \((u_1, v_1, u_2, v_2)\), and lines segments are drawn on the current plot.
- accumulator(skip=1)[source]
Compute the Hough accumulator
- Parameters
skip (int, optional) – increment for line strength threshold, defaults to 1
It creates two new attributes for the instance:
Awhich is the Hough “accumulator” array, rows represent \(\rho\) and columns represent \(\theta\).votesis a list of the number of lines found versus threshold, it can be used to select an optimal threshold.extentis \([\theta_{\mbox{min}}, \theta_{\mbox{max}}, \rho_{\mbox{min}}, \rho_{\mbox{max}}]\).
Warning
The OpenCV
HoughLinesfunction does not expose the accumulator array. This method “reverse engineers” the accumulator array through a costly process of computing the Hough transform for all possible thresholds (increasing in steps ofskip). This is helpful for pedagogy but very inefficient in practice.- Seealso
- plot_accumulator(**kwargs)[source]
Plot the Hough accumulator array
- Parameters
kwargs – options passed to
imshow
The Hough accumulator is computed, if not already existing, and the displayed as an image where brightness is proportional to the number of votes for that \((\theta, \rho)\) coordinate.
- Seealso