*** Welcome to piglix ***

EMVA1288


EMVA1288 is an electronic measurement standard developed by the European Machine Vision Association (EMVA). Its purpose is to define the methods to measure and characterize image sensors and cameras that are used in machine vision. It also provides rules and guidelines on how to report results and how to write device datasheets.

The main goal of the standard is to characterize industrial cameras. Therefore, photography and television standards are not applicable. It was necessary to define a new standard specific to machine vision applications.

The standard is free to use and free to download but the user must register to EMVA to have the right to use the "EMVA1288 compliant" logo on their publications or products.

Work on the 1288 standard started in 2004. Release 1 for monochrome cameras was released in August 2005. In Release A2.01, issued in August 2007 included an additional linearity module. With Release 3, published on November 2010 the first version was available that covered monochrome and color cameras as well as area and line cameras together with a characterization of defect pixels. Release 3.1 came into effect on December 30, 2016. This release contains only a few refinements and additions. Its major new feature is a standardized summary datasheet making camera comparison even easier. The most important refinement is a definition of the camera signal nonlinearity better adapted to cameras with a higher dynamic range. The only two other major additions are: a) the total SNR curve which includes the spatial nonuniformities, and b) diagrams of horizontal and vertical profiles for a meaningful and well-arranged characterization of the different types of the spatial nonuniformities.

The standard only uses radiometric units like watts, joules, number of photons, volts, etc. There is no use of photometric units like lux.

The 1288 standard is based on a linear camera model. All noise sources except for photon noise and quanti­zation noise can be included into a single parameter, the variance of the dark noise. Thus the model contains only three basic unknowns: the quantum efficiency, the dark noise and the system gain.

The standard is made of various modules. Some modules being mandatory, others being optional. For each module, a simple mathematical model of the phenomenon or parameter to be described is built. Then, a method to acquire specific image data is defined. Finally, out of the measured data, the parameter is computed using simple formulas.


...
Wikipedia

...