This is a continuation of a discussion, started in a previous posting, of key factors in successfully building vision machines.

In my experience, the first step in successfully building a vision machine is to learn about the application.  Initially the manufacturing customer knows all about the manufacture, but may know little about machine vision implementation.  The integrator, in addition to knowing about machine vision, has experience in other applications, but may know little about this particular field of manufacturing.


Through communication, laboratory and plant visits, both the customer and integrator learn about the other’s field.


The initial meeting can be either at the integrator’s lab or at the plant.  The requirements, as far as they are presently known, are discussed.

  • Discuss the feature or defect to be detected and what to do with failed parts.
  • Obtain samples, including defects.  This is very important.  Save these carefully, as they are central to the objective specification of the capabilities of the machine.
  • It must be understood that when one is holding a part in the hands and viewing it by eye, one is much more capable of perceiving defects than is a machine, when viewing the part with a camera, while being mechanically handled.  The camera and mechanical handling are much more limited.

If the initial meeting occurs in the integrator’s facility, a plant visit is still necessary in assessing the project.  The points of interest and the dependent critical factors of cleanliness, consistency and clarity, are:

  • see what is the current mode of operation, with regard to (a) contamination (cleanliness), (b) speed of throughput, (c) degree of human interaction possible (with regard to maintenance of cleanliness and consistency of operation),
  • assess current methods of material handling, with regard to (a) positional variation and vibration (consistency), (b) overcoming special characteristics, such as part stickiness (consistency),
  • assess vision implementation for (a) nature of access to needed features (clarity), (b) available space for vision machine (clarity), (c) ability to illuminate (clarity) and (d) presence of glint or other surface characteristics impeding imaging (clarity).

At this point, it is very important to obtain representative samples, both conforming and non-conforming or defective.  Ideally, one should obtain many samples, covering the range of possible defects.


Obtain one of each type of part that will be inspected, particularly if they have novel physical features, such as being made of rubber.


Also, for later performance testing, obtain a large amount of samples, to represent production operation.


This short essay has discussed the collection of project requirements, both through discussions with the prospective client and visits to his or her facility and through the acquisition of actual product samples exhibiting in substantial numerical quantity, both acceptable and unacceptable examples.  In the third part of this essay on successful development of vision machines, I will discuss the need for research, prototypes and mock-ups prior to building the customer’s machine.

Building Successful Vision Machines (Part 2)

Post navigation