ient 
the 
me, 
arse 
line 
ient 
put 
put 
We 
rors 
e is 
ON 
  
137 
S. CONCLUSIONS 
This paper presents a two-stage modular algorithm for precise measurement of the conjuction angle of a welding 
seam. It consists of a detection stage and a measurement stage. 
At the detection stage, the angular object is detected and its parameters are coarsely estimated. This stage consists of 
the Sobel contouring, Hough Transform and Hough accumulator analysis. The special fast Hough Transform with 
natural parametrization is used for computational efficiency and geometrical flexibility. This parametrization permits 
to consider only lines with relatively small negative and positive inclination angles. It also permits to split the initial 
image into a set of vertical stripes in order to analyze a possible curvature of the object lines. The analysis of Hough 
accumulator is realized by the use of its Lateral Histogram. In the case of our test images, such histograms have very 
well expressed the image bimodality, and the histogram modes correspond to the two lines of the angular object. The 
Otsu algorithm is used for detection of these modes. 
At the measurement stage, the Least Squares Method is used for precise measurement of line orientation. The 
selection of contour points is made based on their location in a neighborhood of the line, as well as on the gradient 
direction at these points, which should be approximately perpendicular to the line. The conjunction angle value is 
obtained as the difference between the line slopes. 
REFERENCES 
Otsu N., 1979. "A threshold selection method from gray-level histograms". IEEE Trans. on System, Man, and 
Cybernetics, Vol. 9, No. 1. 
Svalbe I.D., 1989. "Natural representations for straight lines and the Hough transform on discrete arrays". IEEE 
Trans. on Pattern Analysis and Machine Intelligence, Vol. 11, No. 9. 
IAPRS, Vol. 30, Part 5W1, ISPRS Intercommission Workshop “From Pixels to Sequences’, Zurich, March 22-24 1995