Abstract Automatic inspection of PC-boards is one example where distance checking algorithms are needed. The expansion-contraction algorithm due to Ejiri et al. [Computer Graphics Image Processing 2, 1973, 326–329; Elect. Eng. Japan 95, 1975, 90–97] indicates as errors not only widths that are too small but every case where the boundary has a radius of curvature less than half the critical distance d. The C-algorithm, the U-algorithm and the P-algorithm are three new algorithms presented in this paper. These algorithms only indicate an error if the width (= the perpendicular distance form one boundary point to another) is less than the critical distance d. All algorithms have several possible variations but in any case the P-algorithm is advantageous in both cost and speed. An algorithm due to Wojcik is also discussed. All five algorithms as well as a number of variations are compared to each other for different topologies.
[1]
Michihiro Mese,et al.
A process for detecting defects in complicated patterns
,
1973,
Comput. Graph. Image Process..
[2]
Stephen B. Gray,et al.
Local Properties of Binary Images in Two Dimensions
,
1971,
IEEE Transactions on Computers.
[3]
Björn Kruse,et al.
Design and implementation of a picture processor
,
1977
.
[4]
Michihiro Mese,et al.
Defect detection in complicated patterns
,
1975
.
[5]
Zbigniew M. Wojcik.
Automatic detection of semiconductor mask defects
,
1976
.
[6]
Robert C. Restrick.
An Automatic Optical Printed Circuit Inspection System
,
1977,
Optics & Photonics.
[7]
Azriel Rosenfeld,et al.
Connectivity in Digital Pictures
,
1970,
JACM.