When a system designer identifies a need for a component, such as a camera, the first step is of course a website search. The initial search may include basics such as by desired resolution, interface, and frame speed. The next way to narrow down the selection would be by imaging performance criteria such as signal to noise ratio, quantum efficiency, temporal dark noise, and more. The problem was that the values reported from these specifications depended on the testing conditions and the settings can be optimized to provide best possible results.
Over 10 years ago the EMVA started the initiative to define a unified method to measure, compute and present specification parameters and characterization data for cameras and image sensors used for machine vision applications. This standard called EMVA1288 intends to overcome the hurdle of comparing different specification sheets (compare apples to apples).
The specification parameters that are covered are:
- Spectral sensitivity
- Signal/Noise Ratio (maximal SNR, dynamic range, dark noise)
- Inhomogeneities (DSNU, PRNU)
- Defect pixels
For a good explanation about these parameters: http://www.wiley-vch.de/berlin/journals/op/10-01/OP1001_S53-S54.pdf.
The standard has evolved and even more suppliers have continued to adopt the methods for testing. BUT, the testing methods are not used by all suppliers on their data sheets and the EMVA1288 reports are provided separately, partly because the data is complex and difficult to report. It has definitely brought awareness to the challenges with simply comparing data sheets, but it doesn’t mean that the decision on the right camera for an application has become simple. It should be treated as another resource in the image sensor/camera selection process.
For example, the relationship between the parameters measured in the standard and the needs of your specific application still needs to be determined. Even if you determine camera A has better linearity than that for Camera B – what do you require for your measurement? There are of course price consequences for the best performance, so you want to know what you need and not simply which is better in order not to over-design your system. You will likely still need to evaluate several cameras to determine how they behave and perform in your system.
What will EMVA1288 data will provide you:
- The ability to compare data from cameras that have been measured according to the standard
- A shorter list of camera options to select from
- Some camera performance data you can use as input for system performance
What will EMVA1288 data will not:
- Tell you about other artifacts that may be present in the image sensor/camera
- Tell you which camera will work best in your application
- Eliminate the need to evaluate the camera under your system conditions
- Tell you about camera-to-camera performance or camera reliability
- Provide other intangible supplier information such as customization, service, logistics support
The EMVA 1288 standard may help you narrow down your selection, make a comparison on purely camera specs (or image sensor specs most of the time), and understand camera technology better (all great things), but it does not eliminate the need for an in-house evaluation of the camera and a thorough consideration of the company behind the product.