|dc.description.abstract||More than ever, efficiency and quality are key words in modern industry. This situation
enhances the importance of quality control and creates a great demand for cheap and
reliable automatic inspection systems. Taking into account these facts and the demand
for systems able to inspect the final shape of machined parts, we decided to investigate
the viability of automatic model-based inspection of mechanical parts using the dense
range data produced by laser stripers.
Given a part to be inspected and a corresponding model of the part stored in the model
data base, the first step of inspecting the part is the acquisition of data corresponding
to the part, in our case this means the acquisition of a range image of it. In order to
be able to compare the part image and its stored model, it is necessary to align the
model with the range image of the part. This process, called registration, corresponds
to finding the rigid transformation that superposes model and image. After the image
and model are registered, the actual inspection uses the range image to verify if all the
features predicted in the model are present and have the right pose and dimensions.
Therefore, besides the acquisition of range images, the inspection of machined parts
involves three main issues: modelling, registration and inspection diagnosis.
The application, for inspection purposes, of the main representational schemes for
modelling solid objects is discussed and it is suggested the use of EDT models (see
[Zeid 91]). A particular implementation of EDT models is presented.
A novel approach for the verification of tolerances during the inspection is proposed.
The approach allows not only the inspection of the most common tolerances described
in the tolerancing standards, but also the inspection of tolerances defined according to
Requicha's theory of tolerancing (see [Requicha 83]). A model of the sensitivity and
reliability of the inspection process based on the modelling of the errors during the
inspection process is also proposed.
The importance of the accuracy of the registration in different inspections tasks is
discussed. A modified version of the ICP algorithm (see [Besl &; McKay 92]) for the
registration of sculptured surfaces is proposed. The maximum accuracy of the ICP
algorithm, as a function of the sensor errors and the number of matched points, is
A novel method for the measurement and reconstruction of waviness errors on sculp¬
tured surfaces is proposed. The method makes use of the 2D Discrete Fourier Transform
for the detection and reconstruction of the waviness error. A model of the sensitivity
and reliability of the method is proposed.
The application of the methods proposed is illustrated using synthetic and real range