There is a camera. It is necessary to measure its sensitivity.
The matter of measurement of sensitivity is a measurement of scene illumination. The camera image must have Signal/Noise ratio = 17Db. To measure light a Luxmeter is used, to measure Signal/Noise ratio the Image analyzer or the Video analyzer can be used.
The Video analyzer reads out the Signal/noise ratio directly from the screen.
The Image analyzer uses the stored frames in bmp or jpg formats.
In most cases the Video analyzer is more convenient.
If the lower limit of illumination measured by the luxmeter is more than assumed camera sensitivity, see Measuring low illumination.
For measuring IP cameras' sensitivity it is enough to have possibility to display images on the screen.
IP cameras sensitivity can worsen considerably depending on compression level.
If the camera allows to mount only mini-lenses (M12), the mini-lens with a known aperture (F2.0) is needed.
Measuring sensitivity should be carried out in a dark room.
1. Switch off all additional options of image processing. Only AESC and AGC should be switched on. Switch maximal gain of AGC if it is possible.
For IP cameras, set the compression type - M-JPEG, the image quality - the best. While measuring the sensitivity of IP cameras you must know which exposure time is used by the testing camera, otherwise the measurement becomes meaningless.
Sensitivity of IP cameras can depend on the number of pixels in image. The higher the number of pixels is, the smaller pixel size - the worse the sensitivity.
Correctly measuring sensitivity of modern IP cameras is not an easy task. When light is decreased, IP camera automatically turns on the noise reduction, merges neighboring pixels, reduces frame rate, multiplies exposure time, disables color. The black level rises and the noise is lost in black together with dark image details. Meanwhile turning off this automation is impossible with many IP cameras.
2. Mount lens F1.2, on the camera, mount the camera on stand, direct it towards the test chart, connect to computer, display image on the screen.
If the camera allows mounting only mini-lenses (M12), mount a mini-lens with known aperture (F2.0).
3. Direct the camera on the test chart, in the camera field-of-view place the luxmeter sensor.
4. Defocus.the image slightly to blur possible unevenness on the chart.
5. Start Video analyzer, switch to the Signal/noise tab.
6. Click on the Setting sensors checkbox and then mark the darkest and lightest areas by two successive clicks on the image.
At the points of the clicks yellow (dark area) and red (light area) squares will appear, which may disappear after a few seconds when the image changes. Video analyzer will analyze pixels inside these squares (sensors).
The squares must be placed inside even regions with identical brightness. At that, the brightness of these regions should not frequent achieve black and white levels. There should be no roughness in the regions. The roughness will be sensed as noise. Defocus.the image slightly to blur possible unevenness on the chart.
If necessary, re-click on the Setting sensors checkbox and successive specify by clicking the new places for the sensors.
You can change size of the sensors in the Sensors combobox. In most cases, the size of 10x10 (100 pixels) is optimal.
7. To start measurement, click on the Start button. Caption of the button changes to Stop.
Red and yellow squares will appear again at the specified points, and the measurement will start. The measurements will be repeated cyclically every 0,5 seconds.
In the Signal/noise (dB, unweighted) box the calculated Signal/Noise ratio will appear.
About choice of the dark area in details:
The dark area must meet two requirements:
1. Noise at the area should be not be greatly restricted by the black level.
2. It should be exactly the darkest area.
With standard analog cameras it is usually enough to choose the black darkest square on the gray scale. Automation analog camera works in such a way that the black level is kept when changing the light, not cutting off the signal and noise.
Complicated analog cameras with digital processing, IP cameras can behave unpredictably. Sometimes while changing light levels some of squares in the gray scale goes for the black level or rise above the black level, black level "floats." In this case we have to change places for sensors, making sure that the dark area is really darkest and the noise is not cut by the black level.
8. Switch on tungsten halogen lamp. Switch off common light.
9. Changing the distance from the lamp to the sheet of paper, get an image with Signal/Noise ratio = 17dB.
You must not change the lamp supply voltage. It should be equal to nominal lamp supply voltage. Change Illumination only be means of distance variation between the lamp and the sheet of paper.
Be sure, that the luxmeter sensor window and the sensors' places have visually equal illumination.
During reducing light, watch that the noise at the place of sensor on the darkest area is not limited by the black level. The noise is limited when the value in the box Dark area> Brightness is equal to or less than the value in the box Dark area> Noise RMS. If the noise is limited - move the sensor to the darkest place without limitation by the black level.
It is important that at the places of sensors, noise must not be limited by the black level or the white level, otherwise the result of the measurement signal to noise ratio would be inflated.
10. Record illumination according to the luxmeter reading.
If luxmeter sensitivity is not enough, see Measuring low illumination.
11. Value of illumination at which the image is obtained is the Minimum scene illumination for the tested camera at signal/noise ratio equal 17dB and lens F1.2 (F2.0 for a camera with mini-lens).
The Max. level of brightness is the IRE at the minimum illumination.
You can insert these values in VideoCAD into the Table of camera models or in the Sensitivity and Resolution box in VideoCAD. After that it is possible to perform precise modeling of the tested camera model.
With complicated analog cameras and IP cameras may be a situation when we can not get the signal to noise ratio equal to 17dB because of constantly working noise reduction schemes. If the noise reduction can not be disabled, then we can not use the Signal/Noise ratio as a boundary criterion of sensitivity. In this case we have to choose another boundary criteria, for example the image resolution. Precise modeling of the sensitivity of these cameras is not possible, but an approximate simulation of images details may be sufficient for practice.
See also: Illumination and camera sensitivity in CCTV.