Vyacheslav Mikhailovich SMELKOV, PhD in Engineering, Associate Professor
METHOD OF INCREASING THE RESOLUTION OF A TELEVISION CAMERA FOR FORENSIC DIAGNOSTICS
Due to the non-destructive nature of reading information and the convenience of providing it to the consumer, the television method is preferred in conducting various forensic studies.
The most important link in the television hardware designed to solve these problems is the television camera, which (from the standpoint of the information theory of communication, in the information triangle “object – signal – person”), being a source of video signal, can potentially provide the maximum amount of created video information when the speed of its formation is coordinated with the speeds of supply, transmission and reception of information by other components of the television system.
Obviously, the speed of supply of information is determined by the “natural” part of the physical model of the television system, which includes objects of control together with radiation sources or illuminators.
The growth of the amount of information at the output of a television camera is always accompanied by a struggle to increase its “eternal” parameters: sensitivity and resolution.
The latter parameter determines the clarity of the television image formed by the camera.
This article analyzes a method for increasing the resolution of a television camera based on two CCD matrices and operating in a low-frame resolution mode, for which the scanning speed and, accordingly, the image signal rate are significantly slower than in the broadcast standard according to GOST 7845-92.
The low-frame video signal at the camera output has a standard frame format (4/3) and a number of photosensitive elements per line doubled due to electronic “stitching”, providing a twofold increase in horizontal resolution.
The measure of the resolution of a television camera is the number of photosensitive elements (pixels) of the matrix photodetector on the CCD, which performs discretization of the input optical image in two spatial coordinates and time.
The role of the number of pixels in a television image was emphasized by the founder of electronic television V.K. Zvorykin, who formulated the principle of sufficient accuracy in the transmission of video information.
According to this principle, which is also agreed upon by modern television people [1], the number of pixels should be neither too large nor too small.
If the number of pixels is too large and the area of the photo target is limited, the image “drowns” in photon noise, and if it is too small, the number of distinguishable images is sharply reduced.
In the case of distinguishing an object known in advance, the number of pixels per linear size must be no less than a certain number according to the classification, for example, according to the Johnson criterion [2, p. 492].
Unfortunately, the a priori uncertainty for objects observed by a television camera, which is aggravated by the non-invariance of the CCD photodetector response to the shift of the input image, leads to the absence of a universal requirement for the resolution of the television camera.
For forensic diagnostic objects, it is true that they are objects with a high density of individual components, but are in a static state.
For example, such objects are written documents, the study of which primarily concerns their requisites (records, impressions of seals and stamps, typewritten texts, notes, etc.); the materials from which they are made (paper, paints, glues, etc.);
traces left by writing instruments and other devices (seals, stamps, printing machines, etc.);
residues of etching agents used to remove text, etc. [3].
The TV camera must first of all ensure the formation of maximally undistorted video information for its subsequent input into the computer.
It has been theoretically and practically proven in [4, p. 60] that the most advantageous mode of resolution in a CCD TV camera in terms of maximum signal/noise ratio for such objects is the low-frame mode, rather than the broadcast mode.
The use of several CCD photodetectors and electronic “stitching” of images to increase the resolution of a TV camera is encountered in the literature more than once.
Thus, according to the American patent [5], the TV camera uses two CCD matrices with gaps between the pixels. Optical beam separation is performed by an optical block containing a lens and two mirrors, one of which is translucent and the other reflective.
The relative spatial position of the CCD matrices is set in such a way that the pixels of the first CCD matrix are shifted horizontally relative to the second CCD matrix by half the width of one pixel.
This TV camera potentially provides a doubling of horizontal resolution compared to a TV camera with a single CCD matrix, but its practical implementation poses the problem of maintaining high accuracy of the relative displacement of two CCDs.
In addition, the presence of mandatory gaps between elements in the CCD matrices causes losses of light flux, and, consequently, a decrease in the sensitivity of the TV camera.
In another television camera [6], a fiber-optic converter is used to divide the input optical image horizontally into two parts in accordance with the geometric dimensions of the CCD matrix photo target.
The advantage of this method is the elimination of the requirement for gaps between pixels for CCD matrices, but its significant disadvantage is the change in the format of the output image compared to the format of the CCD matrix photo target in a ratio of 2:1
Below is a technical solution for a low-frame television camera based on two CCD matrices.
For this television camera, the number of pixels in the video signal of a line is equal to the total number of elements in the horizontal registers of the CCD, and the frame format coincides with the format of the photo target of a single CCD and is standard (4/3). The structural diagram of the television camera, the device of which is recognized as an invention [7], is shown in Fig. 1.
Fig. 1. Structural diagram of a small-frame television camera according to the invention [7]
It contains an anamorphic lens (1); a beam splitter (2); two CCD matrices (3) and (4); level converters (5), (6), (7), (8), (9) and (10); a master oscillator (11); a sync generator (12); pulse formers (13) and (18); buffer video amplifiers (14) and (15); an adder (16); a divider by two (17).
The essential differences of the television camera are the use of an anamorphic lens and a prism-type beam splitter.
The anamorphic lens is designed to change the scale of the optical image in one direction. In this solution, the anamorphic coefficient of the lens (1) horizontally is 0.5 [8].
The beam splitter (2) ensures that the image from the lens output (1) is divided into two halves with a vertical border, directing them to the photo targets of the CCD matrices.
CCD matrices (3) and (4) are of the same type and can have a circuit organization of “frame transfer” or “line-frame transfer”.
This solution adopts the three-phase organization of “frame transfer”, which is widely used in our country for serial CCD matrices. The frequency of pixel polling in the horizontal registers of both photodetectors is twice as high as the control rate in a television camera on a single CCD.
But since the decomposition mode in the proposed camera is low-frame, this new frequency remains significantly lower than the frequency of element-by-element readout of the CCD when it operates in the broadcast standard. Therefore, this increase in the speed of output of charge elements will not introduce any additional errors into the video signal.
Let's consider the operation of the proposed two-matrix TV camera.
The image of the observed scene is projected through an anamorphic lens (1) and a beam splitter (2) onto the photo targets of the CCD matrices (3) and (4). The anamorphization coefficient of the lens (1) over the frame field is 0.5.
Then the right half of the horizontally stretched optical image of the test object appears on the photo target (3-1) of the first CCD matrix, and the left half of the object image, also horizontally stretched, appears on the photo target of the second matrix.
The master oscillator (11) operates at a frequency twice as high as in a TV camera made on a single CCD matrix. On the other hand, the divider by two (17) ensures the functioning of the synchronizing generator (12) without changing the frequency grid.
In both CCD matrices, at the rate of low-frame decomposition during the forward stroke of the frame scan, the optical images on photo targets (3-1) and (4-1) are converted into charge images, and then the charge packets of frames are synchronously transferred during the reverse stroke to section (3-2) and section (4-2), respectively.
During the forward stroke of the frame scan of the subsequent frame, the charge packets are read one after the other from section (3-2) to the output register (3-3) and from section (4-2) to register (4-3).
In this case, during the first half of the active part of each line of the frame, the charge line of section (3-2) is read at double speed, and during the second half of the active part of the line, the charge line of section (4-2) is read at the same double speed.
Note that the required control of the operation of the CCD registers (3-3) and (4-3) is carried out using blocks (11), (13), (18), as well as level converters (PU9) and (PU10) operating at double the element frequency.
Thanks to this, the video signals at the outputs of both CCD matrices contain video information with a restored (undistorted) scale for the right and left halves of the test object.
Then, through the buffer video amplifiers (14) and (15), the video signals from the CCD are fed to the adder (16), where the procedure of their preliminary processing and “stitching” is performed.
The structural diagram of the adder (16) is shown in Fig. 2.
Fig. 2. Structural diagram of the adder according to the invention [9]
19, 20, 23 – level clamping blocks; 21 – sampling block with “stitching”; 22 – analog multiplier; 24 – buffer video amplifier
The adder (16) performs the following functions:
- restoration of the constant component for the video signal of the first channel;
- restoration of the constant component for the video signal of the second channel;
- selection of sections of video information within each image element in the video signals of each channel and its storage for the period of the element;
- combining (“stitching”) two video signals with the elimination of the effect of the difference in the fixation level;
- equalization of the gain factors of the video signals in each channel and balancing;
- restoration of the constant component for the combined video signal;
- buffer conversion of the combined video signal for operation on a low-impedance load.
The technical solution of the adder (16) can be implemented according to the invention [9]. As a result, at the output “Video” of the adder (16), and consequently of the television camera, a low-frame video signal is formed with the frame format of the CCD photodetector and a doubled number of horizontal decomposition elements.
It is obvious that for the proposed TV camera the horizontal resolution (N x) in TV lines can be determined by the ratio:
N x = (M1 + M2 ) x 1/k,
where
M1, M2 – the number of pixels in a line for the first and second CCD matrices;
k – frame format (4/3).
For modern CCD matrices of typical resolution M1 = M2 = 500. Then N x = 750 TV lines.
If CCD matrices with increased resolution are used as photodetectors, then M1 = M2 = 760. In this case N x = 1140 TV lines.
Literature
1. Nikitin V.V., Tsytsulin A.K. Television in physical protection systems: ETU “LETI”, 2001.
2. Semiconductor image signal generators. Ed. by P. Jespers, F. Van de Wiele, M. White. “Mir”, 1979.
3. Kekin A.G. and [al.]. Hardware for checking the authenticity of documents based on the optical method of non-destructive testing.//Special equipment, 2003, No. 2, pp. 30 – 40.
4. Khromov L.I. and [al.]. Solid-state television. M. “Radio and communication”, 1986.
5. US Patent No. 4,038,690, H04N 3/14, cl. 358/213; 357/24; 357/30. Video signal generation system on a CCD. Published on July 26, 1978.
6. French Application No. 2,476,949, H04N 5/32; H05G 1/60, 1/64. Device for obtaining television images using charge-coupled matrices and a transmitting system with a similar device. Published on August 28, 1981.
7. Decision to issue a Russian Federation patent for an invention under application No. 93047421/09(045521) dated September 21, 1993. MKI6 H04N 3/14, 5/335. Small-frame television camera on charge-coupled devices./V.M. Smelkov. Applicant — FSUE Research Institute of PT «Raster».
8. Cards of anamorphic attachments and blocks for shooting widescreen films./Review of domestic and foreign-made film lenses for shooting conventional, widescreen, wide-format and 16-mm films. Compiled under the editorship of F.S. Novik. Moscow, 1969, p. 165.
9. US Patent No. 4,378,571, H04N 3/14, 5/14, 5/20, cl. 358/213; 358/160. Device for processing an analog video signal of a sequential type for CCD photodetectors. Published 29.03.81.