Research on non-contact measurement based on machine vision

In industrial measurement, due to the limitation of working environment, it is often impossible to carry out manual measurement. Using geometric optics and image processing technology, the target can be measured indirectly through affine calculation. This paper takes the 7mm standard gauge block of national standard level 1 as the target to be measured, compares the measurement results under the environment of front and rear light sources, extracts the edge corners of the target to be measured by sub-pixel corner detection, and proposes a method to measure the target width based on iterative calculation of point line distance. For 7mm target, the calculation result is 7.052mm and the error is 0.74%, which is better than the measurement accuracy based on the minimum circumscribed rectangle.