TechArchive

Original article date: April 1998

The increasing number of industrial applications for machine vision systems are making the technology an integrated part in many production environments. IAN PURCELL, Marketing Manager of Matsushita Automation Controls looks at the different hardware options.

The last ten years have seen a dramatic change in machine vision systems. Much of the development can be attributed to technological advances in CPUs and computer interfaces. Thanks to these advantages and the quality demands of end customers, machine vision systems have become one of the key strategic tools of the production industry.

Today’s machine vision systems offering 100% quality assurance by providing high-speed quality information have now become a cost-effective option for industries involved in mass production. Applications such as fill levels in the food and beverage industry and date codes in the pharmaceutical industry are typical of their use.

Most applications involve a CCD camera, computer based hardware, image processing software and a lighting fixture. The camera, normally positioned above or to the side of the part to be inspected, captures and sends an image at the request of the hardware. The image is stored and analysed to determine its conformance to pre-set criteria and limits. This may be the presence of components within an assembly, the orientation of a part or simply dimensional measurement results.

Serial or parallel data available after each inspection can be used to integrate the vision system into a production line. Two main levels of information are available; the first comprises digital outputs that can be used to activate a reject mechanism and provide simple statistics on the number of good and bad parts over a period of time. The serial alternative can communicate all checking results via a network to a host or other computer based system and can transfer all checking results to allow full statistical process control (SPC) to be employed. In terms of performance, a vision system is much faster and more accurate than a human inspector, and ensures 100% quality assurance.

The advances in computer systems and software have enabled easy to use graphical user interfaces (GUIs) to be designed. These allow an application to be set-up in only a few hours – even by users with little or no previous experience of these systems. In the case of the Matsushita user-interface for the M100 and M200 systems, an eight-button keypad and an on-screen window and menu system is used to configure the systems. Usually, after only a few hours training, the customer can set-up an M100 or M200 system on-line and be able to troubleshoot any problems that may occur during the working life of the vision system.

PC-based machine vision systems using standard operating systems like Windows 95 or Windows NT 4.0 with applications written in Visual C++ or Visual Basic also allow straightforward programming or configuration in the form of a “point-and-click” application programmers’ interface (API). The powerful Matsushita PC based P400 vision system has an API developed in Visual C++ that is user-friendly and allows rapid development of the checking parameters required by the system for the inspection of a part.

There are several common types of machine vision system that combine hardware and software elements.

The Intelligent Camera systems are the simplest and often the most cost-effective. Using a digital signal processor (DSP) and a small CCD sensor, the application is usually developed on a PC and downloaded to the camera unit to allow simple tasks like presence detection and measurements to be performed. Although these systems are fairly compact, all the electronics are built in to the camera unit – as the name suggests. Since they do not have a remote camera option the size can often prohibit installation, especially in a confined area. Also in most cases, since the application for an intelligent camera is developed on a PC, the system does not usually allow the connection of an external monitor that allows online monitoring of the application. This can be an essential feature of an installation allowing an operator to visually monitor the inspection process.

The need for a robust and simple to use ‘intelligent sensor’ led to the development of a type of machine vision system known as “proprietary hardware”. Using specially designed hardware and GUI, this type of system is well suited to the industrial environment. Often in the style and around the same size as a 24 I/O “brick” size PLC, these systems use remote cameras, a remote monitor and are configured using on screen windows and menus via a small keypad.

The M100 and M200 conform to this style and were initially designed to carry out automated inspection functions within Matsushita’s own factories around the world. Modular in design, they have firmware modules that include relevant checkers for certain types of inspection, such as presence checking, measurement, feature extraction and pattern matching.

The use of reduced instruction set computer (RISC) processors and applications-specific integrated circuits (ASICs) allow rapid capture and processing of images in real time. Dedicated digital inputs for sensors to trigger the system when an inspection is required and outputs for strobe light control and communication of pass and fail signals are coupled with a RS232 port that can be used to communicate every result from the inspection routine to a host.

In the last few years, PC based machine vision systems have moved away from highly complicated bundles of various pieces of computer hardware and software that required an understanding of computer science. The new generation of PC systems are now enjoying a rebirth and are a viable option. Generally more expensive, PC-based systems now offer greater flexibility and allow more complex inspections to be carried out. Using a PC as a machine vision system also simplifies connection to a host or communication of results to an SPC package – which may be running on the same PC.

The PC CPU technology, with speeds up to 300Mhz and fast PCI frame grabbers have led to the P400 being developed by Matsushita. An API that uses icons, drop down menus and status displays allows the user to set up the application parameters. Able to perform high level inspections like pattern matching and feature extraction in grey scale, the P400 allows connection of up to 12 cameras and can be networked to other devices to establish full integration in to a process.

Intelligent Camera PC Based Proprietary Hardware

 Table 1 General comparison table of common machine vision system types

April 1998