| Summary: | 碩士 === 國立清華大學 === 工業工程研究所 === 82 === The employment of a machine vision system for part inspection
has become an important component for achieving the goal
of real-time control. However, one of the major
constraints of this technique is its limited accuracy
owing to the pixel resolution. For those applications
which require better accuracy, such as photogrammetry and
precision measurement, the accuracy can be improved by
replacing with better hardware devices or by applying the
subpixel techniques. In this research, we intend to develop
new machine vision techniques to solve the problems
encountered in practical manufacturing environment, with
emphasis on accuracy problem. Linear estimation for LOG
operation (LEFLO) method and weighted average (WA) method are
proposed in this research. The LEFLO method applies the
Laplacian of Gaussian convolution and localizes the edge
with linear interpolation. WA method uses the deviation of
gray values of successive pixels as the weight numbers to
estimate the gray level of the edge translation. In this
research, the synthetic images are employed to evaluate the
effectiveness of the algorithms. The experimental results
show that LEFLO is more accurate than WA when images are
not disturbed by the environmental noise. Nevertheless,
LEFLO is more sensitive to noise. The weighted average
method requires less processing time than LEFLO method.
It is almost one-sixty of the processing time of the LEFLO
method. Finally, an integrated system for socket inspection
is developed to illustrate its capability real-time
inspection.
|
|---|
