TV cameras are imagers using silicon solid state sensors sensitive to visible and near infrared radiation that generate electronic signal in analog video formats or digital video formats that could be treated as successors of classical old television cameras. Most TV cameras are built nowadays using CCD or CMOS technology. Due to historical reasons TV cameras are often called CCD cameras.
From the point of applications TV cameras can be divided into the presented below groups:
- Analog video cameras (general consumer color CCD/CMOS imagers that generate analog video signal)
- Digital video cameras (general consumer color CCD/CMOS imagers that generate digital and analog video signal),
- Television cameras (professional color digital CCD cameras for studio production and electronic news gathering).
- Web cameras (general consumer color CCD/CMOS cameras that generate digital signal in USB format and can be directly connected to PC)
- Network cameras (a version of web cameras for network applications to be used for short range surveillance like CCTV cameras),
- CCTV cameras (monochrome/color CCD/CMOS imagers that generate typically analog video signal for short range surveillance applications in indoor/outdoor conditions),
- Day/night surveillance CCD cameras (color/monochrome CCD/CMOS imagers that generate analog/digital high quality video image). These cameras can be treated as improved, more expensive version of CCTV cameras. Very good sensitivity is achieved due to advanced electronic processing systems used in these cameras. The cameras are typically equipped with large zoom objectives.
- ICCD cameras for night applications (high sensitivity cameras built by coupling CCD/CMOS sensor with image intensifier tube).
- EBAPS cameras (surveillance high sensitivity cameras based on novel CMOS sensors built using Electron Bombarded Active Pixel Sensor technology – coupling GaAs photocathode with CMOS sensor),
- Cooled CCD cameras (high sensitivity cameras built using cooled CCD/CMOS sensors – mostly used in astronomical or scientific applications)
- Linear imagers (imagers built using linear high resolution silicon detectors – mostly for space or airborne applications)
- Digital still cameras (CCD/CMOS based equivalent of classical photo cameras).
Here we are interested in testing and evaluation of TV cameras used in surveillance applications. The cameras from group 5-9 can be potentially used as surveillance TV cameras. The majority of surveillance TV cameras are low cost CCTV/network cameras designed for short range observation. Dozens of test methods were proposed for testing these cameras. As CCTV/network cameras are low cost devices then the proposed test solutions do not require hi-tech expensive test equipment. We are however interested in testing surveillance TV cameras designed for long range observation, quite often for observation at day/nigh conditions. Here we are to test quite expensive and important modules of surveillance systems and professional test equipment is needed for testing these cameras. From now we will understand the term surveillance TV cameras as high-end cameras from group 6-9 with high quality zoom objectives designed for medium/long range observation.
Fig. 1. Two exemplary surveillance TV camera
There are some standards that regulate testing low cost surveillance CCTV cameras used in security applications for short range observation. There are standards that regulate testing TV cameras used in machine vision. There is a similar situation with testing of digital still cameras. However, there is no internationally accepted standard that regulates testing and evaluation of surveillance TV cameras designed for medium/long range observation where the most important evaluation criterion is detection, recognition and identification range of target of interest.
In situation of a certain vacuum of legal regulation we use proposed at the end of 1990s (Holst Gerald, CCD Arrays, Cameras, and Displays, JCD Publishing, 1998) a popular concept to extend methodology of the thermal imaging systems performance model FLIR 92 to TV cameras, or wider to all visible imaging systems. If we accept this concept then we can characterize TV cameras using slightly modified parameters, test methods, evaluation methods of thermal imagers. In the latter field testing and evaluation is quite well developed and normalized.
Comparison of parameters of thermal imagers and parameters of surveillance TV cameras
|Thermal imagers||TV cameras|
|Minimal Resolvable Temperature Difference||Minimal Resolvable Contrast, resolution|
|Modulation Transfer Function||Modulation Transfer Function|
|Field Of View||Field Of View|
|Noise parameters(NETD, FPN, non uniformity, 1/f, dead pixels)||Noise parameters(NETD, FPN, non uniformity, 1/f, dead pixels)|
|Responsivity function(responsivity (SiTF), linearity, dynamic, saturation level)||Responsivity function(responsivity (SiTF), linearity, dynamic, saturation level)|
|3D Noise model||3D Noise model|
|Other (PVS, SRF, ATF)||Other (PVS, SRF, ATF)|
As we see in Tab. 1 TV cameras can be characterized using similar set of parameters as used to characterize thermal imagers. There are only several differences. First, Minimum Resolvable Contrast characteristic becomes the most important parameter of surveillance TV cameras. Second, Noise Equivalent Input becomes a measure of high frequency temporal noise in TV cameras. Three, two new parameters like sensitivity and SNR are added in case of TV cameras as parameters of such names are used for a long time to characterize these cameras.
The tests of TV cameras can be carried out at three stages of development:
- detection sensor: CCD/CMOS sensor, TDI sensor, ICCD module, EBAPS sensor
- TV camera module (TV camera without optics)
- complete TV camera
Characteristics of TV cameras
|Parameter||Complete TV camera||TV camera module|
|Minimal Resolvable Contrast||x||x|
|Modulation Transfer Function||x||x|
|Noise parameters (Noise Equivalent Input, 1/f noise, Fixed Pattern Noise, non-uniformity, SNR, SiTF, dead pixels)||x||x|
|Responsivity function (responsivity, linearity, dynamic, light range||x||x|
|3D noise model||x||x|
Resolution is typically defined in military standards as maximal spatial frequency of a standard USAF 1951 target that can be revolved by an observer at a certain light level and the target contrast
Minimum Resolvable Contrast function is a function of a minimum contrast difference between the bars of the standard target and the background required to resolve the image of the bars by an observer versus spatial frequency of the target at different levels of target luminance. The USAF 1951 target is used as the standard target during MRC measurement.
In other words MRC is at a resolution measured at different illumination levels for different contrast levels; and the resolution is a point at the MRC characteristic. Having known MRC function it is possible to calculate the ranges of detection, recognition and identification of the target of interest.
MTF is a module from the OTF function of the imaging systems, where OTF is Fourier transform from the image of the point source. In other words, MTF is defined as an output signal modulation to input signal modulation when the signal is sinusoidal wave. It is a measure of the degradation of an output image as correlated to the input pattern which is normalized to 100 percent contrast at zero spatial frequency.
Sensitivity is a minimal illumination value required to produce 50% of the nominal signal at the camera output.
Different values of minimal required illumination level are used in different sensitivity definitions.
Conditions for sensitivity measurement should be always clearly stated: if the illumination was measured at sensor plane or target plane, color temperature of the light source, reflectance of the target chart, parameters of the optics, what gain was set, what gamma was used, if other correctors were used etc.
Noise present in signal generated by tested TV imager can be described using different parameters. Here we assume that the noise can be characterized by a set of four parameters:
1/f – a measure of low frequency temporal noise
Fixed Pattern Noise: a measure of high frequency spatial noise
Non Uniformity: a measure of low frequency spatial noise.
The presented above set of four noise parameters gives users of TV cameras precision information about influence of noise on overall image quality. In case of designers of TV cameras mode detail information is needed. It is then recommended to use so called 3D noise model where noise is characterized by nine parameters.
Signal to Noise Ratio S/N is a parameter defined as a ratio of the nominal signal at the camera output to the true root-mean-square (rms) value of the signal fluctuations.
It is recommended to measure two types of noise: spatial noise (fixed pattern noise) and temporal noise (random noise). Therefore two types of SNR are later calculated. In both cases measurement of noise should be carried out at illumination level that produce 50% of nominal output.
Responsivity function is a function that present relationship between output signal generated by tested TV camera (typically in digital levels) on input light intensity (in luminance units or illuminance units). This relationship depends on camera settings: gain, brightness, optics aperture, time exposure, optic aperture etc. Responsivity function is usually S shaped.
Several more detailed parameters can be determined on the basic of measured responsivity function or several responsivity functions).
Signal Transfer Function (SiTF) or responsivity is the quasi linear part of responsivity function.
Saturation Level – illumination level when output signal is saturated.
Dynamic is a ratio of the illumination that produce nominal signal at camera output measured for minimal sensitivity settings of the TV camera (minimal optical aperture, minimal gain, maximal shutter speed) to minimal illumination that can produce half of the nominal signal at camera output measured at maximal sensitivity settings of the camera (maximal optical aperture, maximal gain, minimal shutter speed).
Light Range: the range between the illumination that produce nominal signal at camera output measured for minimal sensitivity settings of the TV camera (minimal optical aperture, minimal gain, maximal shutter speed) and the minimal illumination that can produce half of the nominal signal at camera output measured at maximal sensitivity settings of the camera (maximal optical aperture, maximal gain, minimal shutter speed).
Color fidelity is defined as color difference between color of original target and color of its image (parameter applicable only to color TV cameras). Color fidelity is measured using software analysis of images generated by tested TV camera of a reference standard color target under variable illumination conditions.