LSD

class cbclib.bin.LSD(scale: float = 0.9, sigma_scale: float = 0.9, log_eps: float = 0.0, ang_th: float = 45.0, density_th: float = 0.7, quant: float = 0.02)

LSD is a class for performing the streak detection on digital images with Line Segment Detector algorithm [LSD].

Parameters

• scale (float) – When different from 1.0, LSD will scale the input image by ‘scale’ factor by Gaussian filtering, before detecting line segments.

• sigma_scale (float) – When scale is different from 1.0, the sigma of the Gaussian filter is sigma = sigma_scale / scale, if scale is less than 1.0, and sigma = sigma_scale otherwise.

• log_eps (float) –

  Detection threshold, accept if -log10(NFA) > log_eps. The larger the value, the more strict the detector is, and will result in less detections. The value -log10(NFA) is equivalent but more intuitive than NFA:

  • -1.0 gives an average of 10 false detections on noise.

  • 0.0 gives an average of 1 false detections on noise.

  • 1.0 gives an average of 0.1 false detections on nose.

  • 2.0 gives an average of 0.01 false detections on noise.

• ang_th (float) – Gradient angle tolerance in the region growing algorithm, in degrees.

• density_th (float) – Minimal proportion of ‘supporting’ points in a rectangle.

• quant (float) – Bound to the quantization error on the gradient norm. Example: if gray levels are quantized to integer steps, the gradient (computed by finite differences) error due to quantization will be bounded by 2.0, as the worst case is when the error are 1 and -1, that gives an error of 2.0.

References

LSD

“LSD: a Line Segment Detector” by Rafael Grompone von Gioi, Jeremie Jakubowicz, Jean-Michel Morel, and Gregory Randall, Image Processing On Line, 2012, DOI: 10.5201/ipol.2012.gjmr-lsd, http://dx.doi.org/10.5201/ipol.2012.gjmr-lsd.

scale

sigma_scale

log_eps

ang_th

quant

detect(image: numpy.ndarray, cutoff: float, filter_threshold: float = 0.0, group_threshold: float = 1.0, dilation: float = 0.0, profile: str = 'linear', return_labels: bool = False, num_threads: int = 1)

Perform the LSD streak detection on an input array image. The Streak detection comprises three steps: an initial LSD detection of lines, a grouping of the detected lines and merging, if the normalized cross-correlation value if higher than the group_threshold, discarding the lines with a 0-order image moment lower than filter_threshold.

Parameters

• image (numpy.ndarray) – 2D array of the digital image.

• cutoff (float) – Distance cut-off value for lines grouping in pixels.

• filter_threshold (float) – Filtering threshold. A line is discarded if the 0-order image moment is lower than filter_threshold.

• group_threshold (float) – Grouping threshold. The lines are merged if the cross-correlation value of a pair of lines is higher than group_threshold.

• dilation (float) – Line mask dilation value in pixels.

• profile (str) –

  Line width profiles. The following keyword values are allowed:

  • tophat : Top-hat (rectangular) function profile.

  • linear : Linear (triangular) function profile.

  • quad : Quadratic (parabola) function profile.

  • gauss : Gaussian funtion profile.

• return_labels (bool) – Return line labels mask if True.

• num_threads (int) – A number of threads used in the computations.

Returns

A dictionary with detection results. The dictionary contains a list of the following attributes:

• lines : An array of the detected lines. Each line is comprised of 7 parameters as follows:

  • [x0, y0], [x1, y1] : The coordinates of the line’s ends.

  • width : Line’s width.

  • p : Angle precision [0, 1] given by angle tolerance over 180 degree.

  • -log10(NFA) : Number of false alarms.

• labels : Image where each pixel indicates the line segment to which it belongs. Unused pixels have the value 0, while the used ones have the number of the line segment, numbered in the same order as in lines.

Return type

Dict[str, Dict[int, numpy.ndarray]]