THE “PIR” METHOD FOR A 24-HOUR SECURITY CAMERA..
SMELKOV Vyacheslav Mikhailovich
THE “PiP” METHOD FOR A 24-HOUR SECURITY CAMERA
The term “PiP” is an English abbreviation for the phrase “picture in picture”, which means “image in image” or “picture in picture”.
Looking back at the “PiP” method in television, it should be noted that it was first proposed for use in television receivers, giving the viewer the opportunity to watch two or more programs on the screen at the same time [1].
Currently, the “PiP” function is a mandatory mode in modern television models with a screen diagonal of 29 inches and larger.
The author of this article proposes to implement the “PiP” method in a security camera in order to increase the probability of the operator making the right decision due to high-speed scaling within the boundaries of the television image window of the violation zone.
When modeling security television systems (STS), it is considered a postulate that a lot of time spent on checking the truth of the alarm signal and assessing the situation negates all efforts to protect objects [ 2] . Therefore, within a fairly short period of response time, the STS operator must make the right decision.
It is known that the operator's decision-making process includes four interrelated stages: detection, classification, discrimination and recognition. Detection means distinguishing an object from the background and assigning it to classes of objects of potential interest. Classification means assigning the detected object to one of the broad classes (person, vehicle). Distinction means assigning the observed object to a narrower subclass (truck, passenger car). Finally, at the recognition stage, the type of object (car brand) is established.
For example, according to [http://cctv-information.co.uk], the UK Home Office recommends having at least 10% of the raster for the height of the image of the person being detected, and at least 50% for person identification.
Considering that modern zoom lenses have at least a 6-fold range of focal length change, a simple way to ensure both detection and identification of an object with one camera is to use a zoom lens in it.
However, it should be agreed that the finite time for changing (adjusting) the focal length of the zoom lens is in conflict with the required short response time of the SOT operator to an alarm situation.
The paper [3] proposes a patented US technical solution for a TV camera on a matrix of devices with charge coupling (CCD), providing selective electronic enlargement of a section (fragment) of an image and its placement in a combined image at the place of the initial selection. In terms of hardware capabilities, the equivalent of this TV camera is two conventional ones plus a video mixer. The implementation method of “PiP” is digital, due to which the scaling time is practically eliminated, and the enlarged image of the intruder in the window can be shown to the operator immediately (without delay) after the alarm is registered. Since the operator is simultaneously presented with the image of the violation zone and its surroundings, the probability of a correct decision is additionally increased.
But this solution is accompanied by a significant drawback, consisting in the reduced resolution of the enlarged fragment both in the longitudinal and transverse directions in relation to the resolution indicator of the rest of the image. This is determined by the fact that in the enlarged image of the selected area, compared to its original (normal) image, the growth of the geometric dimensions of the fragment is accompanied by the use of the same number of light-sensitive elements of the matrix CCD to form both images. Thus, the enlarged image has a reduced value of the frequency-contrast characteristic (MCF) for the middle and upper spatial frequencies, and the operator perceives it as stretched in both directions.
Therefore, the 6-fold digital scaling of the image in the window of this camera does not allow the operator to move from the “human detection” level of vision to the “human recognition” level of vision.
Specialists from the Federal State Unitary Enterprise Research Institute of Industrial Television “Raster” (Veliky Novgorod) have developed technical solutions for television cameras with selective scaling [4, 5], which implement the high-speed analog “PiP” method, but without loss of image resolution.
In development of these solutions, a structural diagram of a new television camera for round-the-clock security is proposed, based on two (color and black-and-white) CCD matrices. The structural diagram of the camera is shown in Fig. 1.
Fig. 1. Structural diagram of the “day-night” television camera
The technical result of the proposed solution is as follows:
- high-speed automatic formation by the camera during the day, upon an alarm signal, of a combined image, which consists of a monochrome window containing a black-and-white image of the central fragment of the violation zone, enlarged in scale, and a color image on the rest of the raster in normal scale (the scale of the initially presented image) with an unchanged resolution index within the entire combined image;
- implementation of automatic “day-night” switching according to the criterion of the a priori selected signal-to-noise ratio with the transition of the camera from the color image mode during the day to the black-and-white image mode enlarged in scale for the entire raster at night.
The television camera contains an optical unit (1), a first television signal sensor (2), a second television signal sensor (3), a switching and shaping unit (4), a sync pulse selector (5), a peak detector (6), a motion detector (7), a recording and reset pulse generator (8), a sampling and storage unit (9), and a comparator (10).
The optical unit (1) contains a beam splitter (1-1), a correcting filter (1-2), the first objective (1-3) and the second objective (1-4). The beam splitter (1-1) is made in the form of a prism with two beam-splitting faces located at an angle of 30°, and the correcting filter (1-2) is made in the form of a prism with one beam-splitting face. The ratio of the focal length of the second objective (1-4) to the focal length of the first objective (1-3) determines the scaling factor (Km) of the optical unit.
Let's distinguish three modes in the operation of the TV camera:
- “Day” (mode 1);
- “Night” (mode 2);
- “Alarm” (mode 3, which can accompany both mode 1 and mode 2).
Regardless of the camera operating mode, the optical image of the test object along the optical path: the first beam-splitting facet of the beam splitter prism (1-1), the second beam-splitting facet of the beam splitter prism (1-1), the spectrum-splitting facet of the prism of the correcting light filter (1-2), the second facet of the prism of the correcting light filter (1-2), the first objective (1-3) is projected in the visible spectral range onto the photo target of the sensor (2). At the same time, the central fragment of this image, magnified by a factor of Km, along another optical path: the first beam-splitting facet of the beam splitter prism (1-1), the second beam-splitting facet of the beam splitter prism (1-1), the third facet of the beam splitter prism (1-1), the second objective (1-4) in the entire spectral range (visible and infrared) is projected onto the photo target of the second sensor (3).
Note that the infrared region of the spectrum of the last image is additionally enhanced by the light flux reflected by the spectrum-splitting face of the prism of the correcting light filter in the direction of the third face of the beam splitter.
As a result of photoelectric transformations, a color signal of the image of the test object is formed at the output of the first sensor (2), and a black-and-white signal of its enlarged (in accordance with the scaling factor) central fragment is formed at the output of the second sensor (3), which act as parallel full television signals. The selector (5) extracts line and frame sync pulses from the full television signal of the sensor (2).
Regardless of the operating mode of the television system, the generator (8) generates the following recording and reset pulses with a period of half frames (Tp) within each frame blank.
The peak detector (6) with the period Tp measures the video signal level from the sensor (2), the sample-and-hold block (9) registers this level and remembers it for the half-frame time, and the comparator (10) evaluates the output voltage of the block (9), comparing it with the threshold voltage Uп.
Let's assume that the TV camera operates in mode 1, and the comparator (10) does not change its output state, maintaining the state of logical «1». The presence of logical «1» at the second control input of the block (4) ensures the transmission of the full television signal from the sensor output (2) to its output.
Let the motion detector (7) be prepared for operation, i.e. a protected zone is selected (dotted line along the perimeter in Fig. 2a), which marks the zone of increased interest for the operator on the image.
Let us assume that the motion detector (7) is then initialized, i.e. set to the mode of analyzing changes in the input video signal in the selected zone. Let us assume that in mode 1 there are currently no moving objects (intruders) in the protected zone. Therefore, at the second output of the motion detector (7) there is a logical “1”, which, when fed to the first control input of the unit (4), blocks (cancels) the formation of a monochrome window at its output. A “normal” in scale color image of the space observed by the first sensor (2) is reproduced on the screen of the video monitoring unit (Fig. 2a). Then, let an intruder appear in the protected zone at some point. Then the motion detector (7) will register changes in the video signal, and at its second output the logical “0” level will be set. The TV camera switches to the “1+3” mode. As a result, a combined image is formed at the output of the unit (4), containing a “normal” color image and a monochrome window, in which a black-and-white image of the violation zone is transmitted with a linear magnification of Km times. This combined image signal is fed in transit through the motion detector (7) to the input of the video control unit for presentation to the operator, as shown in Fig. 2b. It should be noted that the monochrome window with a linear increase in content appears without delay, i.e. almost instantly in relation to the moment of automatic registration of the intruder, which is extremely valuable for security television systems.
a) in mode “1”;
b) in mode “1 + 3”
Fig. 2. Characteristic images formed by the television camera:
After the operator has made a decision on the fact that the camera has registered a violation in the protected area, it must be switched back to mode 1. To do this, the operator forcibly resets the alarm signal by acting on the second input of the detector (7).
If the illumination at the facility decreases significantly, for example, when switching from daytime to evening, then the range of the video signal of the first sensor (2) will also decrease.
If the value of the video signal level measured at time t0 by the peak detector (6) and recorded by the sample-and-hold block (9) is less than the threshold voltage Uп, then the comparator (10) will go into the logical “0” state, and the camera will go into mode 2.
Then, internal switching will be performed in block (4), and the full television signal from the sensor output (3) will be transmitted to its output. As a result, a black-and-white image of the protected area, “wide open” to the entire raster, with a scaling factor of Km will be reproduced on the screen of the video control block. It should be noted that the nature of this image will not change if a violation occurs in the protected area, which will be registered by the motion detector (7), and the television camera switches to the “2+3” mode.
If in this situation after the forced reset of the alarm the illumination at the facility increases as a result of the daily transition “night-day”, then the proposed camera will automatically switch from mode 2 to mode 1 again.
Conclusion
Undoubtedly, the development of SOT is moving towards the intellectualization of technology based on digital methods of video signal processing. However, the operation of security systems shows that the widespread introduction of computers does not exclude the role of humans in decision-making due to the great uncertainty of scenes at protected facilities and the irreversibility of processes generated by erroneous decisions.
The author hopes that the implementation of the analog method for generating alarm images proposed in this paper can really help the security operator in his work.
References
- UK Patent for Invention No. 1495173. MKI2 H04N 5/44. Published on 14.12.1977.
- Nikitin V.V., Tsytsulin A.K. Television in Physical Protection Systems. Publishing House of St. Petersburg ETU «LETI», 2001.
- US Patent for Invention No. 4002824. MKI2 H04N 7/00. Published on 11.01.1977.
- Television Camera with Selective Scaling. Patent for invention of the Russian Federation No. 2171014. MKI7 H04N 5/225, . 5/228 Authors: Smelkov V. M., Smolyakov Yu. A., Egorova N. N., Petrova V. M. Published on 20.07.2001.
- Television camera with selective scaling. Patent for invention of the Russian Federation No. 2174745. MKI7 H04N 7/00, . 7/18 Authors: Smelkov V. M., Smolyakov Yu. A., Antonov V. E., Petrova V. M. Published on 10.10.2001.