December 20199MANUFACTURINGTECHNOLOGYINSIGHTSthe best ways to build and deploy a successful vision application. Edge, blob, or histogram vision tools are extremely useful because they are fast and can easily be made binary. They are easily setup and maintained. Growing and adapting vision systems are easily performed with smart cameras. Because of this, it is important to ask a follow-up question, "What is the potential scope creep?" Scope creep refers to changes or uncontrolled growth in a project's definition, after the project's final design approval has been defined. Scope creep can be created from variables such as the change of a supplier or an additional inspection the customer might want to add. Smart cameras are scalable. Smart cameras are easily modified with different lens or lighting. This is not limited to just 2D camera. Today 3D vision has become more versatile and economical. Some systems have even developed configuration wizards that will walk the user through the setup. The idea is always to keep it simple and robust. This holds true even for legacy vision systems already existing on a machine. I once experienced a customer who wanted to add a new camera and another inspection to an existing machine. As the discussion about the scope of the inspection progressed, we discovered the simplest solution was to just add another light to the existing inspection and use the light only when certain product is running. In another case, a customer wanted to install a 3D vision system to check if a box was empty. While the 3D system would work well, the simpler vision solution was to use a low-resolution camera to see the dark parts in the white box. This inspection merely required a basic histogram analysis tool. A practical vision engineer looks for the simplest solution that exists within the scope of the project. Tier 2 solution looks for the opportunity to integrate a simple vision application using simple vision algorithm tools to successfully and robustly solve the vision system application.Tier 3­Advance Vision SystemsThe third question a practical vision engineer asks is, "Am I in over my head?" Most of the tier 1 and 2 applications can be easily and successfully deployed with common vision knowledge. Tier 3 applications, however, require expert level knowledge and expertise in algorithms, technology and integration. Applications including scopes requiring high-level precision, extensive inspection and/or high speed are in this tier. In addition, applications that require rigid quality control specification also could fall into tier 3. This applies both to 2D and 3D vision systems. Solutions in tier 3 do not necessarily have to be PC-based. Smart cameras, in many cases, can give highly accurate and repeatable results. It is all about the science required in selecting the hardware components and applying very deliberate vision algorithm tools to successfully solution the application requirements. One concern with such elaborate vision systems is that they will be difficult to maintain. That is not necessarily the case. A robust vision system simply is well planned to consider and address all the challenges possible before the building process begins. Therefore, it is imperative to be crystal-clear on the scope of the application before reaching outside of your organization for help. A practical vision engineer will work closely with the selected integrator to build a successful application. Tier 3 solution utilizes the experts in advance vision systems. A practical vision engineer is someone who asks the questions, the right questions, and all of the questions. Striving to understand the scope of the given application and build the most successful solution. The solution might be using basic sensors technology, simple vision integration, or advance vision systems. Lean and effective solutions are the path to becoming a successful and practical vision engineer. One must approach each application as a practical vision engineer. Practical vision engineers take time to understand the scope of the application and ask three fundamental questions that lead them to choose one of three tiers of machine vision solutionsAlex Reznichenko
< Page 8 | Page 10 >