DUAL-MATRIX TV CAMERA FOR SURVEILLANCE IN DIFFICULT LIGHTING CONDITIONS.

DUAL-MATRIX TELEVISION CAMERA FOR SURVEILLANCE IN DIFFICULT LIGHTING CONDITIONS..

Vyacheslav Mikhailovich SMELKOV, PhD in Engineering, Associate Professor

Source: Special Equipment magazine, No. 2, 2006.

The first issue of the Special Equipment magazine for 2006 published an article entitled “Possibilities of Building a Security Television Camera for Observation in Complex Lighting Conditions.” In it, with regard to operational features associated with camera operation in the presence of “backlight” or in a “against the light” situation, a method for combating accompanying video signal distortions aimed at expanding the dynamic range was proposed. The implementation of this method envisaged the implementation of a television camera based on two synchronously and in-phase matrices of charge-coupled devices (CCD). A priori, it was considered possible to preliminarily perform spatial orientation of the television camera so that strongly illuminated objects would be perceived in the central part of its field of view angle.

According to the mentioned method, the video signal was generated for each of the two sensors using automatic sensitivity adjustment (ASA) based on the charge relief of a separate area of ​​the photo target, with the first sensor implementing this function for the central area of ​​the photo target, and the second sensor for the entire area of ​​the photo target except for the central area. As a result, the primary generation of the components of the combined image was performed completely automatically with different values ​​of the adjustable parameters (exposure time and video path gain) for each of the transmitted fragments of the scene.

However, if the illumination of the controlled central fragment of the limiting dimensions becomes excessively high under the observation conditions, this may lead to the component of the charge signal of the white spot going beyond the central photometric area and the occurrence of “edge leaks” in the image of the second sensor. Let us assume that the illumination of the entire remaining peripheral observation area, on the contrary, is extremely low. Then these factors together may cause significant distortions of the corresponding part of the combined image due to non-optimal automatic control parameters set in the second sensor.

Below is a technical solution to the problem that arises in the semi-automatic mode of operation of the device. The structural diagram of the television camera is shown in Fig. 1. It contains a sequentially located and optically connected lens 1 and a beam splitter 2, a first television signal sensor 3, a second television signal sensor 4, a sync selector 5, a “window” signal generator 6, a switch-mixer 7, a peak detector 8, a comparator 9, a switch 11 and an RS trigger 12.

In the proposed solution, television signal sensors 3and 4, as before, are synchronized in Genlock mode with frequency and phase binding to the receiver synchronization signal (RSS) from sensor 4.

Fig. 1. Structural diagram of the TV camera

The VSI-746 frameless camera offered by the Russian company EVS can be used as sensor 4, and 3 can be used as sensor– VNI-702 frameless camera [1], both of which are based on a CCD matrix with 582×752 elements and a target diagonal size of 1/2 inch. Information exchange between the sensors is also possible, but then the SSP pulses must be fed from sensor 3 to the “sync” input of sensor 4. Note that the AFC photometry in these sensors is performed over the entire area of ​​the photo target.

A special feature of this solution is that sensor 4 has the first and second control inputs.

For the VSI-746 device, the first control input is pin 20 of the CXD2463R microcircuit of the synchronizing generator. If it is necessary to turn on the AFC by accumulation time (ATS), it is necessary to apply a logical “0” to this pin, to switch to the manual control mode – a logical “1” in TTL levels.

The second control input of the VSI-746 device is formed by pins 11, 12, 13 of the same CXD2463R microcircuit. For operation in the ARVN mode, these pins are “hanging in the air”, since the corresponding potentials in the range of 1.3 – 3.5 V are supplied to them using high-resistance resistive dividers. For manual control of the photodetector accumulation time, it is possible to switch eight values ​​of fixed exposures in the range from 10 μs to 8.33 ms. The necessary code combinations of zeros and ones, which must be supplied to the corresponding pins, are given in Table. 1.

Table 1

These code combinations can be executed using a three-digit counter 10. The electrical circuit of this counter is shown in Fig. 2. Output “A” of the counter is connected to pin 11 of the CXD2463R microcircuit, and pins “B” and “C” are connected to pins 13 and 12 of this microcircuit, respectively.

Fig. 2. Electrical circuit of the counter

Counter 10is a counter of a non-cyclic type, but a self-stopping one. The counter circuit is based on the technical solution proposed in [2, pp. 172–173]. The counter contains the first (T1), second (T2), and third (T3) JK-type triggers, an AND element, an OR-NOT element, and a NOT element.

Counter 10 starts counting from the binary number “000”, which is guaranteed by briefly applying a logical “0” signal to the trigger clearing inputs. Then comes the number “100”, then “010”, etc., as shown in Table 1.

If during the counting process an external signal of logical “1” appears at the first input of the “OR-NOT” element, then a signal of logical “0” will be generated at the J- and K-inputs of the first trigger, which will lead to blocking and stopping the counter.

If during the counting process the signal of logical “0” is held at the first input of the “OR-NOT” element, then upon reaching the number “111” the signal of logical “1” will be set at the output of the “AND” element. This signal, fed to the second input of the “OR-NOT” element, will set a signal of logical “0” at its output. As a result, the counter will also be stopped.

Shaper 6is designed to obtain a “window” signal with the format (AxB) at the output,

whereA is the size of the “window” in the raster horizontally;
B is the size of the “window” in the raster vertically.

Within the raster, the “window” occupies the central fragment, and its dimensions are related to the raster dimensions by the following ratios:

A = X/(2…3),
B = Y/(2…3),

where X and Y are the horizontal and vertical dimensions of the raster, respectively.

It is advisable to generate the “window” signal using a digital method, for example, based on the PIC16C73-201/SP processor, which is widely used in Russia.

The peak detector 8 is designed to store the voltage proportional to the maximum level of the video signal, which is generated by the second sensor 4, in the frame area located outside the “window”. The peculiarity of the peak detector 8is memorized only if there is a high logic level at its strobe input. Before the start of the next operating cycle, the detector is reset using a positive pulse supplied to the “reset” input.

Comparator 9 is designed to compare the level of the information signal from the output of the peak detector 8 and the threshold voltage Uп with a stepwise change in the output voltage in the case when the information signal is greater than Uп.

Switch 11ensures that when a logical “1” is applied to its control input, the binary number signals from the output of the counter 10 bits are connected to the second control input of the first sensor 3. When a logical “0” is present at the control input of the switch 11, the second control input of the sensor 3 is isolated from the counter 10.

Block 12 is a logical trigger device of the RS type with a high active level at the control inputs.

The TV camera (Fig. 1) operates as follows.

Let the camera's field of view simultaneously contain both brightly and poorly illuminated objects and/or objects with a sharp difference in brightness. The camera is preliminarily oriented so that brightly illuminated or bright objects are perceived in the central part of its angle of view.

As in the previous solution, the input optical image passes along the optical path: lens 1, beam splitter input 2, the first beam splitter output 2 is projected onto the photo target of the first sensor 3television signal. At the same time, this image passes along another optical path: lens 1, beam splitter input 2, the second beam splitter output 2 is projected onto the photo target of the second television signal sensor 4.

As a result of photoelectric conversion, the optical image of each sensor is further converted into the corresponding video signals, and from the full television signal formed at the output of sensor 4, selector 5selects horizontal and vertical sync pulses. At the output of the generator 6 a pulse signal of the “window” of positive polarity is generated, which ensures at the output of the switch-mixer 7 the formation of a complete television signal of a combined image, consisting of a video signal from the sensor 3 within the boundaries of the “window” and a video signal from the sensor 4 in its remaining part.

It should be added that the automatic adjustment of the accumulation time (ATT) of the photodetectors for the sensor 3, and for sensor 4 will set practically the same value of the current exposure in both channels for a strongly illuminated or bright scene. But due to the small and non-optimal value of the accumulation time of the photodetector of sensor 4, this will lead to an inevitable limitation of the dynamic range of brightness gradations for the objects of control transmitted in the combined image outside the “window”.

To eliminate this drawback, a positive polarity pulse is fed to the “start” input of the TV camera. During the pulse action, the counter 10 is cleared, which forms the number “000” at the output of the digits, and the detector 8 is reset.

A logical “1” signal is set at the direct output of the RS trigger 12. The latter is fed to the control input of the switch 11 and to the first control input of the sensor 4. Therefore, the ARVN circuit in the sensor 4is switched off, and its second control input is connected to the output of the counter 10. The accumulation time of the photodetector of sensor 4 is set to 10 μs (Table 1).

Counter 10 performs a direct count of frame sync pulses, and with each change in the output number, the accumulation time of the photodetector of sensor 4 increases sequentially. Therefore, the level of the video signal generated by sensor 4 for dark and/or low-light objects increases.

Peak detector 8regularly (with a half-frame period) registers an increase in the video signal, and the comparator 9 compares this count with the threshold voltage Uп.

Let us assume that at some point the output voltage of the peak detector 8 reaches the value Uп. Then the comparator 9 flips over, and a logical “1” signal is set at its output. As a result, the counter 10 stops, and a binary number is recorded at the second control input of the sensor 4, determining the value of the accumulation time of the photodetector.

This camera mode is semi-automatic, since it involves adjusting the response threshold (Uп) of the comparator 9, in order to achieve a compromise result between increasing the signal/noise ratio for the peripheral area of ​​the combined image and the edge distortions that occur at the boundaries of its central area. To increase the accuracy of this adjustment, it is recommended to switch the clock input of the counter 10 at this time.for an increased period of counting pulses taken from the frequency divider output, as shown in Fig. 2 by dotted lines.

It should be added that if, upon reaching the maximum number (“111”) at the output of the counter 10, the voltage at the output of the peak detector 8 does not reach the value Uп, the counter will stop on its own. In this case, the accumulation time of the photodetector of the sensor 4 will be 8330.0 μs (Table 1).

To return the camera to automatic operation mode, it is necessary to apply a negative polarity pulse to the “stop” input. Then a logical “0” signal will be set at the direct output of the RS trigger 12, and the operation of the ARVN circuit will be restored in the sensor 4.

With the same geometric dimensions of the photo targets of sensors 3 and 4, the components of the combined image (in the “window” and outside the “window”) will have an unchanged scale.

If it is necessary to have an enlarged image within the “window”, the geometric dimensions of the photo target of the second sensor must exceed the corresponding dimensions of the first sensor. Let the diagonal size of the target for the first sensor be 1/4 inch, and for the second — 1/2 inch. Then the scaling factor of the combined image will be 1/2 : 1/4 = 2 times.

1. TV cameras of the company «EVS»: Catalog, 2005.
2. Tokheim R. Fundamentals of digital electronics. /Translated from English. Moscow: Mir, 1988.