Real resolution of a television camera

Real resolution of a television camera

A.N. Kulikov.

REAL RESOLUTION OF A TELEVISION CAMERA

Source — Magazine «Special Equipment» №2 2002

When designing a security television system, the resolution of the television camera stated in the passport is usually taken into account.

Based on this, the surveillance zones and installation locations of television cameras are determined, the angles of the field of view are calculated and the lenses are selected.

Then multiplexers, video recorders and other devices are purchased.

The installers lay cables, install television cameras and equipment, and finally the system is turned on.

At first glance, everything works fine, the monitors show images of the premises and territories of the facility.

But at the very first incident it turns out that the offender's face is impossible to discern. The number of the entering car is not visible, and sometimes it is even impossible to discern its make.

In the dark, things are even worse: images of parts are blurred, moving objects are smeared.

As a result, the television system, instead of full-fledged surveillance, provides the security service with functions close to the capabilities of conventional security sensors.

This is due to the fact that the actual resolution of television cameras and its dependence on illumination, depth of field, as well as the loss of resolution in the cable network, multiplexers, video recorders and other devices are not taken into account when designing the system.

The article examines the factors influencing the resolution of a television camera operating as part of a security television system.

Resolution of a television camera and the number of photodetector elements.

The parameter “resolution” came to television from optics.

Initially, the limit of resolution, according to the Rayleigh criterion, was understood to be the distance between two points at which the center of one spot coincides with the middle of the first dark diffraction ring of the second spot (Fig. 1)

Fig. 1 Resolution of an optical system.

E max, Emin – illumination of the light and dark diffraction rings, respectively,
D is the diameter of the entrance pupil,
f‘ is the back focal length,
d is the linear resolution limit,
l is the wavelength of light.

In this case, the relative difference in illumination at two adjacent points (the depth of signal modulation at the frequency of maximum resolution) is approximately equal to 26% of the maximum illumination [ 1 ].

With the advent of discrete photodetectors (CCD matrices), the concept of optical resolution has become imprecise due to the appearance of the effect of superposition of spatial frequencies of the test target lines and the photosensitive elements of the matrix.

Nevertheless, the parameter resolution is used in advertising brochures for television cameras.

It should be noted that the resolution of a discrete photodetector depends on the position of the test target lines relative to the grid of elements of the photosensitive matrix.

a) the centers of the strokes coincide with the centers of the image elements,

b) the centers of the strokes are shifted by half the size of the element.

Fig. 2 Illustration of the change in the maximum resolution
of a discrete photodetector when shifting it
relative to the image of the target by 1/2 the element size.

It is evident (Fig. 2) that in the case when the number of target lines is equal to the number of photodetector elements along the measured coordinate, there can be two extreme values ​​of resolution.

If the target lines fall exactly in the center of the CCD matrix elements, then the resolution at the camera output will be maximum, and a thin grid will be visible on the video monitor.

If you shift the target by half a line, then the maximums and minimums of the line image will fall in the middle between the CCD elements and each element will have a half signal (average between black and white) and the monitor screen will only have a flat gray background.

If the number of horizontal target lines is less or greater than the number of matrix elements, a flat gray background will also be observed when shifting the target position, but not on the entire image, but in the form of separate vertical columns (moire).

With a decrease in the number of lines of the target, the visibility of moires will decrease, however, even with half their number, relative to the number of CCD elements, they will still be quite visible (Fig. 3).

Fig. 3 Illustration of the image of moires of the vertical wedge
of the test table, observed by a television camera on the CCD matrix.
Below is an oscillogram of a line in the center of the horizontal target of 450 — 600 television lines.
Moiré is expressed in low-frequency modulation of the oscillogram.

In order to match the parameter resolution with the number of elements of the CCD matrix along a given coordinate, it was proposed to determine the resolution by multiplying the number of elements by a coefficient of 0.75.

Currently, the most common CCD matrices in security television cameras are of two types: standard and high resolution, with the number of elements per line of 500 and 750, respectively (Currently, the latest television cameras for security systems are beginning to use “megapixel” CCD matrices, similar to the matrices of digital cameras.

The resolution of such cameras with the number of elements per line of about 1600 exceeds 1000 television lines.).

Multiplying by 0.75, we get approximately 380 and 560 television lines for standard and high resolution television cameras.

At first, TV camera manufacturers indicated these values ​​in their data sheets.

Unfortunately, some companies, for advertising purposes, try to increase the generally accepted coefficient and indicate the resolution for their cameras as 420 and 600 lines, although they use the same CCD matrices with 500 and 750 elements, respectively.

 An undocumented parameter of TV cameras is the depth of signal modulation at the maximum resolution frequency.

Comparing cameras made on the same CCD matrices, one can see that, despite the declared identical resolution, the clarity of the images they form is different.

Some cameras, even made on high-resolution matrices, have a blurry, “muddy” image, while other cameras, on the contrary, pleasantly surprise with the filigree rendering of small details.

Nevertheless, formally, the resolution of the camera that forms a fuzzy image corresponds to the value specified in the passport.

If you look closely at the image of the vertical wedge of the test chart formed by this camera, then with difficulty, but still, you can see the 560 lines stated in the passport.

In “sharp” cameras, these lines are visible without difficulty, they are well “drawn” and have high contrast.

Why is there such a difference in clarity in cameras with the same CCD matrices?

The fact is that the image is projected onto the CCD matrix by lenses whose characteristics are close to the limit. This is due to the very small size of the photosensitive cells of modern matrices.

For example, the size of the element of the ? inch CCD matrix, high resolution ICX-209AL from SONY is 4.85 x 4.65 µm, which is only a few times greater than the diffraction limit at the long-wavelength boundary of the matrix's spectral range (Fig. 1).

In addition, chromatic aberrations and lens manufacturing inaccuracies mean that the circle of confusion of modern lenses often exceeds the geometric size of the matrix element.

This means that the frequency-contrast characteristic of a television camera will have a noticeable drop, starting from half the resolution of the CCD matrix, and at the frequency of the maximum resolution, the contrast often does not exceed 10% compared to the image contrast in a large detail (Fig. 4).

Fig. 4 Frequency-contrast characteristic of a high-resolution television camera
with the sharpness corrector turned off.

If measures are not taken to correct the frequency-contrast characteristic of the lens in a television camera, the resulting image will be blurry, which can often be observed in cheap cameras of Eastern assembly.

In higher-class cameras, special clarity correctors are installed to compensate for losses in the lens. Correctors come in different types.

In a simple case (for example, the WAT-902H camera from WATEC), an asymmetric corrector is installed that emphasizes the first derivative of the signal.

The best results are obtained by symmetrical adaptive sharpness correctors that take into account the second derivative of the signal, the degree of correction of which depends on the illumination of the image (camera VNC-742 from EVS).

To assess the real sharpness of the image, the parameter “modulation depth of the signal at the frequency of the maximum resolution” is used, equal to the ratio of the signal amplitudes from the target with the number of strokes equal to the maximum resolution, and with the minimum number of strokes (large image detail).

It is evident (Fig. 5) that the signal amplitude at a frequency of 550 lines in a camera with a symmetrical sharpness corrector significantly exceeds these values ​​in cameras with a corrector by the first derivative and even more so in a camera without a sharpness corrector.

a) – CV-300 television camera without a sharpness corrector

b) – television camera with asymmetrical sharpness corrector WAT-902H

c) – a television camera with an adaptive, symmetrical clarity corrector VNC-742.

Fig. 5 Images (top) and oscillograms of a line of 550 television lines (bottom) of a vertical wedge,
obtained using three different high-resolution television cameras,
with identical TO412FICS lenses installed in them at an aperture value of F 8.0.

Unfortunately, the parameter modulation depth (in some sources called “amplitude of frequency-contrast characteristic at the frequency of maximum resolution”) is not given in advertising brochures and passports for television cameras.

Therefore, the real resolution of a television camera can only be assessed by observing the image formed by the camera during its testing.

Loss of resolution and depth of field in lenses with automatic diaphragm.

To expand the range of working illumination of television cameras, lenses with automatic diaphragm control (ARD) are installed in them.

When using such lenses, it is possible to obtain a range of working illumination from 0.01 lux to 100,000 lux and even wider, that is, to ensure the operation of the camera both day and night [ 2 ].

Particularly popular at present are the so-called “aspherical” lenses with a minimum relative aperture of up to 0.75.

However, from the point of view of resolution, when using ARD lenses, a number of unpleasant moments arise:

• The depth of signal modulation at high spatial frequencies in ARD lenses depends on the aperture value, and with a fully open aperture can decrease by 10 or more times.
• The depth of field (the range of distances within which a given image clarity is ensured) depends to an even greater extent on the aperture value, and is minimal when the aperture is fully open.
• Light scattering in the lens [ 2 ] also depends on the aperture value and is maximal when the aperture is fully open.

Consequently, the resolution and contrast of the image of a television camera with an ARD lens are significantly reduced in the evening, and especially at night, when the lens aperture is fully open (Fig. 6).

Fig. 6. Dependence of the signal modulation depth (amplitude of the frequency-contrast characteristic) on the horizontal target of 550 TVL
on the relative aperture of the lens (aperture value) in a high-resolution television camera,
with a TO412FICS lens from Computar installed.

a) The value at F 0.8 was obtained with the aspherical lens HG0608AFCS-HSP from the same company installed.

Images of the central part of the test chart, formed by a high-resolution television camera with a lens with relative apertures installed:
b) F 0.8
c) F 2.0
d) F 8.0

In addition to the general deterioration in clarity, at night there is also additional defocusing of objects at different distances, images that were clearly focused during the day.

Defocusing occurs not only due to the decrease in depth of field with a fully open aperture, but also due to changes in the spectral composition of the light source (the Sun or artificial lighting). Particularly strong defocusing occurs at night when using IR floodlights.

This leads to two rules that must be followed when installing cameras with ARD lenses:

• Cameras with ARD lenses must be focused in the dark, when the lens aperture is fully open (depth of field is minimal), and the appropriate artificial lighting is turned on.
• In cameras with ARD lenses, it is necessary to turn off the built-in electronic shutter system, otherwise, the lens aperture will be fully open not only at night, but also during the day, with the resulting loss of resolution and depth of field.

At night, with insufficient artificial lighting, the main reason for the loss of camera resolution will be the influence of the television camera's own fluctuation noise [ 3 ].

The camera's resolution begins to deteriorate sharply with a decrease in the signal-to-noise ratio.

When the signal-to-noise ratio decreases from 40 dB (100 times) to 20 dB (10 times), which is usually the threshold sensitivity of a television camera, the resolution decreases from 500 to 100 television lines (Fig. 7).

a) with a signal-to-noise ratio of 40 dB

b) and with a signal-to-noise ratio of 20 dB.

Fig. 7. Illustration of a decrease in resolution
when a television camera observes text with different font sizes

A special case of loss of resolution occurs in television cameras with long-focus lenses (25 mm and longer) designed to observe distant or extended objects.

These losses are due to several factors.

First, the circle of confusion of a real lens increases with focal length, starting at about 16 mm (for cameras with a CCD matrix format of 1/3 inch and less).

Secondly, when observing at long distances, air turbulence has a noticeable effect, especially if there are open windows of warm rooms, heating pipes, running motors of mechanisms or other warm objects near the camera along its axis of vision.