Skip to contents

Line Segment Detection in an Image

Source: R/utils_imagem.R
image_line_segment.Rd

Detects line segments in a digital image using the Line Segment Detector (LSD), a linear-time method that controls false detections and requires no parameter tuning. Based on Burns, Hanson, and Riseman's method with an a-contrario validation approach.

Usage

image_line_segment(
  img,
  index = "GRAY",
  scale = 0.8,
  sigma_scale = 0.6,
  quant = 2,
  ang_th = 22.5,
  log_eps = 0,
  density_th = 0.7,
  n_bins = 1024,
  union = FALSE,
  union_min_length = 5,
  union_max_distance = 5,
  union_ang_th = 7,
  union_use_NFA = FALSE,
  union_log_eps = 0
)

Arguments

img

An Image object.

index

A character string with the index to be used. Defaults to "GRAY".

scale

A positive numeric value. Scales the input image before detection using Gaussian filtering. A value <1 downscales, >1 upscales. Default is 0.8.

sigma_scale

A positive numeric value determining the Gaussian filter sigma. If scale <1, sigma = sigma_scale / scale; otherwise, sigma = sigma_scale. Default is 0.6.

quant

A positive numeric value controlling gradient quantization error. Default is 2.0.

ang_th

A numeric value (0-180) defining the gradient angle tolerance in degrees. Default is 22.5.

log_eps

A numeric detection threshold. Larger values make detection stricter. Default is 0.0.

density_th

A numeric value (0-1) defining the minimum proportion of supporting points in a rectangle. Default is 0.7.

n_bins

A positive integer specifying the number of bins for pseudo-ordering gradient modulus. Default is 1024.

union

Logical. If TRUE, merges close line segments. Default is FALSE.

union_min_length

Numeric. Minimum segment length to merge. Default is 5.

union_max_distance

Numeric. Maximum distance between segments to merge. Default is 5.

union_ang_th

Numeric. Angle threshold for merging segments. Default is 7.

union_use_NFA

Logical. If TRUE, uses NFA in merging. Default is FALSE.

union_log_eps

Numeric. Detection threshold for merging. Default is 0.0.

Value

A list of class lsd containing:

  • n - Number of detected line segments.

  • lines - A matrix with detected segments (columns: x1, y1, x2, y2, width, p, -log_nfa).

  • pixels - A matrix assigning each pixel to a detected segment (0 = unused pixels).

References

Grompone von Gioi, R., Jakubowicz, J., Morel, J.-M., & Randall, G. (2010). LSD: A Fast Line Segment Detector with a False Detection Control. IEEE Transactions on Pattern Analysis and Machine Intelligence, 32(4), 722-732.doi:10.5201/ipol.2012.gjmr-lsd

Examples

library(pliman)