EMVA 1288 — Machine Vision Test Standard

Standard: EMVA 1288 — Standard for Measurement and Presentation of Specifications for Machine Vision Sensors and Cameras.
Technical Committee: European Machine Vision Association
Published: 2016-12-30

 

This page is under development. This message will be remove when it’s ready.

The EMVA 1288 standard characterizes image sensor quality (but not full camera systems, which include lenses). Imatest will implement a subset of the standard in the Imatest 2020.2 release, which will be available in late 2020 (and available sooner in the Imatest Pilot program).

Most of the new measurements are derived from two sets of flat-field images, read into Uniformity or Uniformity Interactive using signal averaging to separate temporal noise from spatial nonuniformities.

In Imatest 2020.2, we will will only measure parameters that can be derived from images. We will not (for now) work with photons or absolute light values, so we will not refer to η (quantum efficiency; the number of electrons per photon; \(\eta = \mu_e/\mu_p\)) or K (the overall system gain with units of DN/electron, where DN is the “digital number”, i.e., pixel level). Since input images have a variety of bit depths and hence maximum values (e.g., 255 for 8-bit images; 65535 for 16-bit images, etc.), internal calculations are double precision normalized to a maximum of 1. We will generally used these numbers where DN is called for.

Results are derived from two sets of averaged images read, where each set has a minimum of 4 images (with 8 or more strongly recommended).

  1. Light images, nominally at the 50% level for linear images (about 73% for gamma = 2.2 (standard color space) images).
  2. Dark images (“lens cap on”)

Here are some of the abbreviations used in the standard.

  • μ is the mean of a signal. μd is the mean of the dark signal (dependent on exposure time and temperature). 
  • σ2 is the variance of a signal (related to noise power, which is additive for uncorrelated noise sources). Noise has two components, which need to be separated: noise nonuniformity and temporal noise. 

The key results calculated by Imatest are

     
     
     
     
     
     

Related Measurements Currently Available

Metric Imatest Module Description
Raw sensor Dynamic Range MultichartsMultitest Fits raw data to an equation from the EMVA 1288 standard, then extrapolates to find DR. The test chart does not have to have as large a tonal range as the DR, but transmissive charts with tonal range ≥ 3 are recommended.
Hot & Dark Pixels Uniformity & Blemish detect defective pixel clusters
Photo Response Nonuniformity (PRNU) Uniformity & Blemish map uniform light field uniformity

 

Related Measurements Currently Available

Metric Imatest Module Description
Raw sensor Dynamic Range MultichartsMultitest Fits raw data to an equation from the EMVA 1288 standard, then extrapolates to find DR. The test chart does not have to have as large a tonal range as the DR, but transmissive charts with tonal range ≥ 3 are recommended.
Hot & Dark Pixels Uniformity & Blemish Detect defective pixel clusters