Line detection in gray-scale images is a low-level processing which has applications in, for example, determining markings and roads, etc. in GIS map images. One traditional line detection technique utilizes template matching. In this paper we present a new method which is based on adaptation of the notion of wavelet-based edge detection at coarser scale levels and localization of center line pixels between the two parallel edges. The wavelet decomposition provides smoothed images and detail images at successive resolutions from which multi-scale edges can be obtained. A line (curve) becomes thicker at coarser scales after smoothing which provides two narrowly separated parallel strings of wavelet maxima around it; the center line (curve) between the two can be detected as the line of interest. When lines representing different markings with different gray values intersect, the method will be able to detect individual lines.
[1]
Truong Q. Nguyen,et al.
Wavelets and filter banks
,
1996
.
[2]
Stéphane Mallat,et al.
Characterization of Signals from Multiscale Edges
,
2011,
IEEE Trans. Pattern Anal. Mach. Intell..
[3]
Charles K. Chui,et al.
An Introduction to Wavelets
,
1992
.
[4]
E. L. Usery.
A feature-based geographic information system model
,
1996
.
[5]
Liang-Hwei Lee,et al.
Vision-Based Image Processing of Digitized Cadastral Maps
,
1996
.
[6]
Muhittin Gökmen,et al.
Edge Detection and Surface Reconstruction Using Refined Regularization
,
1993,
IEEE Trans. Pattern Anal. Mach. Intell..
[7]
Ching-Chung Li,et al.
Nonorthogonal wavelet edge detector with four filter-coefficients
,
1993,
Optics & Photonics.