Measurement of Structure’s Image Primitive Based on Power-Law
In industrial machine vision (MV) applications, it is a popular research direction to identify a structure mechanical part image primitive’s relationship with its background by extracting actual contour from a captured image. However, numerous nonlinear problems turn up in MV applications of measuring and inspecting mechanical parts. Those problems are brought by components of MV system including lens, cameras, lighting effect, and sheltered measured object. These nonlinear phenomena usually result in incomplete and inconsistent geometrical shape of a structure’s image primitive. Therefore, special primitive measurement theory and method are researched in the thesis, on the basis of analyzing fractal and structural features of mechanical parts. The method found that an important trait of mechanical parts image contour which can be used to reduce the impact of nonlinear phenomena and determine a closest contour of the image primitive by using different multi-scale rulers in measuring. The effect of the measurement method is proved by an example in cylindrical connectors’ inspection.