How do you think the new GigE standards will influence the machine vision industry?
Respond or ask your question now!
Much has been written about the relative advantages of CCD and CMOS technologies for image sensors. While such discussions are academically interesting, they reflect an us-versus-them approach more suited to promoting a given technology than to helping solve a machine vision design problem. Looking first at the application will allow developers to match each technology's strengths to their needs.
The first application detail to consider is the type of image sensor needed. Image sensors for machine vision are typically one of two types: area sensors or line sensors. An area sensor is a two-dimensional array of pixels that captures an image of an entire region all at once, like a snapshot. A line sensor is a one-dimensional, linear array of pixels that machine vision systems can use to build a 2D image one line at a time.
There also are variations of the line sensor that are important enough to merit individual attention. One is the dual-line sensor that, as its name suggests, has two parallel linear arrays. Having the parallel pixel array gives the dual-line sensor added sensitivity.
The other important line-sensor variant is the time delay and integration (TDI) sensor. This sensor has multiple linear arrays in parallel but is not an area sensor. Instead of exposing and transferring data from all the arrays simultaneously, the TDI sensor scans across the arrays. The timing of this scan coincides with the movement of an image segment across the array so that the signals from each array reflect the same image segment. The result is data that represent only a single line of the image, but are the accumulation of the information from all of the parallel arrays.
Choosing Sensor Type
The various sensor types address different applications. Area sensors are usable for most machine vision applications, but are required when the objects being studied are moving in two dimensions. Examples include security systems, where the scene may be stationary but the objects of interest (i.e., intruders) are moving throughout the scene, and robotic vision, where the system must control the 2D movement of an arm. Area sensors also are used for making detailed inspections of a random area within a larger object.