RU2407028C2 - Device for detecting thermal objects on background of celestial hemisphere - Google Patents

Abstract

FIELD: physics.

SUBSTANCE: in the method, a target image is focused on a radiation receiver, amplitude of the picked up signal is converted to a digital code and stored in form of a two-dimensional array U_N,M=||u_n,m||, where N is number of rows and M is number of columns of the array. The element of the array un,m is proportional to brightness of background radiation picked up from the cell of the receiver in the n-th row and m-th column. The modulus of the threshold value for the decision making interval is determined. The array U_N,M is transformed into an array E_n,m with elements e_n,m and an array

is formed: if e_n,m lies in the interval [-p, p]> then

=0 s and if e_n,m lies outside the interval [- p, p], then

=1. The rectangular monitor screen is divided into cells according to the number of rows and columns of the array

. Cells

=1 are allocated colour having the highest contrast relative the background monitor screen, and considered marks from point thermal objects.

EFFECT: improved method owing to automatic and adaptive accounting for the level of radiance and spatial dimensions of radiation of a small thermal object and cold extended atmospheric background.

4 dwg, 12 tbl

Description

The invention relates to methods for processing an optical image obtained by an optoelectronic system (OES) for locating point thermal objects (heat direction finders) operating on an atmospheric background in the infrared wavelength range. The considered heat direction finders have an optical system with a wide field of view, a short focal length and a matrix receiver of infrared radiation. They are attributed to the survey ECO. The angular size of the observed point object in them is less than or equal to the elementary angular field of the matrix infrared (IR) receivers [1, p. 58].

A method for detecting point thermal objects [2, p. 64], based on the use of a threshold when separating marks from a thermal object and background taking into account the average value of background radiation, is known for such heat finders. A point object (target) is understood to mean a small-sized thermal object, the image of which fits in the elementary field of view (image pixel) of the direction finder or at the same time falls into several adjacent vertical or horizontal pixels.

According to the method [2], the optical system of the direction finder is guided and then fixed in the segment of the celestial hemisphere of the search, equal to the angular size of the field of view of the optical system of the direction finder, in which it is assumed to be a thermal object. The phono-target image is focused on the sensitive areas of the matrix multi-element radiation detector 1 (Fig. 1) and, using the horizontal reading scheme, an electrical signal is proportional to the two-dimensional distribution of the energy brightness of the background radiation and the thermal object. The amplitudes of the signal are converted into a digital code using an analog-to-digital Converter 2 (figure 1). The digital code is stored in the operational memory of the video processor in the form of a two-dimensional array U _{N, M} = || u _{n, m} ||, where N is the number of rows and M is the number of columns of the array, respectively, so the element of this array u _{n, m} contains information on the voltage proportional to the brightness of the background radiation taken from the cell of the multi-element receiver in the nth line at the mth step.

Next, the elements of the array U _{N, M are} converted into video signals, and on the monitor screen of the video receiver, a phono-target image is formed in the form of brightness marks, the column and row numbers of which correspond to the column and row numbers of the array U _{N, M.}

To detect a thermal object, a threshold algorithm is used. The threshold is selected taking into account the fact that the brightness of the radiation of the “cold” atmospheric background and the thermal object differ in magnitude. The response threshold is visually selected according to the average brightness level of the background noise. Those areas of the image where the pixel brightness is higher than the set threshold are considered a thermal object (target), and the rest are considered a background. It is assumed that the two-dimensional brightness field of a random background has a normal distribution law [2, p. 65].

The disadvantages of the method include the need to obtain advance information about the statistical characteristics of the background to select a detection threshold. This approach is practically unrealizable [2, p.65]. This is due to the fact that the correlation radius of atmospheric backgrounds averages 50 s [3, p.135]. This means that the lifetime of a background with certain statistical characteristics is limited. Therefore, it is necessary to change the threshold for automatic decision making with a certain frequency to maintain a given probability of detection. In addition, it is known that the distribution law of most atmospheric backgrounds is bimodal, i.e. different from normal [3, p.141]. Therefore, when using automatic detection, taking into account the average level of the background, the probability of making erroneous decisions increases.

The objective of the invention is to develop a more advanced amplitude method for detecting thermal objects, automatically and adaptively taking into account the differences in energy brightness levels and spatial (angular) sizes of radiation of a small-sized thermal object and a cold extended atmospheric background.

To solve the problem of the invention, it is proposed to use a method through periodic compensation of the background. To implement such compensation, it is proposed to create an array through periodical compensation (IPPC), the values of the elements of which are calculated by the formula w _{n, m} = u _{n, m} × (-1) ^{(n + m)} , where u _{n, m are the} elements of the image array U _{N, M} , which must have an even number of N rows and an even number of M columns. If extended interference enters the angular field of view of the device at any moment of time, the number of irradiated elements along the rows and columns of the MCF W _{n, m} giving a signal of positive polarity will be approximately equal to the number of elements giving a signal of negative polarity. When summing the amplitudes of the bipolar signals recorded along the rows or columns of the MPCH, their mutual reduction will occur, and the arithmetic average of the rows or columns containing signals from the atmospheric background will be close to zero. When averaging bipolar values of elementary signals over the row or column of the MCF containing the amplitude of the signal from a thermal object, mutual compensation will not occur, because the radiation brightness of a “cold” atmospheric background and a thermal object always differ in amplitude [1, p.20]. As a result, the arithmetic mean value in such rows and columns will be nonzero.

Thus, it is possible, by setting the decision threshold for the arithmetic average calculated from the rows and columns of the MPCH, to determine the number of the row and column in which the amplitude of the signal from the object is recorded. The proposed method includes simple mathematical operations and can be implemented in real time in a typical video processor [1, p.225].

The structural diagram of the heat finder that implements the proposed method is presented in figure 1. The proposed method for detecting thermal objects due to periodic compensation of the background includes the following operations.

1. Obtaining an image array U _{N, M} from an analog-to-digital converter of video processor 2 (Fig. 1) of size N × M, where N is an even number of rows and M is an even number of columns of an array.

2. The calculation of the average brightness of the background noise of the image array U _{N, M} 3 (figure 1) according to the formula

where u _{n, m} is an element of the array U _{N, M} containing information about the signal proportional to the brightness of the background radiation taken from the cell of the multi-element receiver in the nth line at the mth step.

3. The determination of the module of the threshold value p for the decision interval 4 (figure 1) by the formula

where k is the weight coefficient that allows you to adapt the decision threshold.

4. The formation of IPPC 5 (Fig. 1) W _{N, M} = || u _{n, m} ||, in each element having indices n and m, which is recorded the result of multiplication in accordance with the formula

where u _{n, m} elements of the image array U _{N, M} , which must necessarily have an even number of N rows and an even number of M columns.

5. The calculation of the average values of the elements of the array W _{N, M} in each of its row 6 (figure 1) by the formula

, где элементы вектора являются средними значениями, рассчитанными для соответствующих строк МЧПК 7 (фиг.1).6. Vector formation

where the elements of the vector are the average values calculated for the corresponding lines of the IPPC 7 (figure 1).

7. The calculation of the average values of the elements of the array W _{N, M} in each of its column 8 (figure 1), according to the formula

, где элементы вектора являются средними значениями, рассчитанными для соответствующих столбцов 9 (фиг.1).8. Vector formation

where the elements of the vector are the average values calculated for the corresponding columns 9 (figure 1).

9. The formation of a two-dimensional array E _{N, M} 10 (Fig. 1), in each element with indices n and m which record the multiplication result in accordance with the formula e _{n, m} = µ1 _n · µ2 _m , where µ1 _n and µ2 _m are the values of the nth and mth elements of the vectors M1 and M2, respectively.

=1, а если значение величины элемента e_n,m массива E_N,M находится в пределах интервала [-p, p], то величину элемента обнуляют:

=0.10. Threshold processing 11 (Fig. 1), during which for all elements of the array E _{N, M} check the fulfillment of the double inequality -p≤e _{n, m} ≤p. If the value of the element e _{n, m of the} array E _{N, M} goes beyond the interval [-p, p], then a unit value is assigned to it:

= 1, and if the value of the element value e _{n, m of the} array E _{N, M} is within the interval [-p, p], then the element value is reset to zero:

= 0.

, для чего прямоугольный экран монитора 13 (фиг.1) видеоконтрольного устройства делят на прямоугольные ячейки по числу строк и столбцов массива

, ячейки

, содержащие единичные значения, выделяют цветом, имеющим наибольшую контрастность по отношению к фону экрана монитора. Принимают решение, что в элементах массива со значениями

=1 находятся отметки от точечных тепловых объектов. По номерам строк и столбцов элементов, отличающихся по цвету от фона экрана монитора, определяют пространственное положение тепловых объектов в сегменте полусферы поиска. При отсутствии элементов

c единичными значениями принимают решение об отсутствии тепловых объектов в анализируемом сегменте небесной полусферы поиска.11. The formation of the phono-target image on the monitor of the video receiving device 12 (figure 1) using the values of the elements of the array

why the rectangular screen of the monitor 13 (figure 1) of the video monitoring device is divided into rectangular cells by the number of rows and columns of the array

cells

containing unit values are highlighted with the color that has the highest contrast with respect to the background of the monitor screen. Make a decision that in array elements with values

= 1 are marks from point thermal objects. By the numbers of rows and columns of elements that differ in color from the background of the monitor screen, the spatial position of thermal objects in the segment of the search hemisphere is determined. In the absence of elements

with single values decide on the absence of thermal objects in the analyzed segment of the celestial hemisphere of the search.

The flowchart of the method of inter-period background compensation is presented in figure 2.

The essence of the proposed method lies in the fact that the radiation of a point thermal object and atmospheric background differ not only in level, but also in angular dimensions. The correlation radius of the cloudy sky usually does not exceed an angle of 1-3 ° [3, p.129]. This indicates that the inhomogeneities of the extended cold atmospheric background (1-3 °) are large in comparison with the small size of the radiating point thermal object. The elementary field of view corresponding to one pixel (with a matrix size of 256 × 256) is 3.5 arc seconds, i.e. the image of the target is several hundred times smaller than the images of inhomogeneities of the atmospheric background.

To compensate for interfering reflections, it is proposed to use the property of atmospheric background radiation, which has approximately the same level of brightness fluctuations within a hemisphere segment with angular sizes of 1-3 ° [3, p. 129]. The size of the field of view of the heat finder is 18 ° in elevation and 18 ° in azimuth. The size of the MPC is 256 × 256 elements, i.e. when translating it into a visual image, it will contain 256 × 256 = 65536 pixels. This makes it possible to distinguish 42 segments of the angular field of view of a device measuring 2 ° 15 '× 2 ° 15', which corresponds to the division of the MPD into fragments of 32 × 32 elements. Thus, the MBMS can be divided into 42 or less fragments, within which there is an equal number of irradiated elements of positive and negative polarity. The random placement in the detection segment of regions with different background radiation intensities, as well as their different spatial extent, does not significantly affect the result of averaging radiation in a particular column (row). Incomplete compensation of the background radiation may be due to an odd number of elements belonging to the private background radiation region, which differs in level of radiation from other private regions. Nevertheless, if an extended interference gets into the segment of the angular field of view at any moment of time, the number of irradiated elements along the rows and columns of the MBPC W _{n, m} giving a signal of positive polarity will be approximately equal to the number of elements giving a signal of negative polarity. When summing the amplitudes of the bipolar signals recorded over the rows or columns of the IPPC matrix, they will cancel each other, and the arithmetic mean value over the rows or columns containing signals from the atmospheric background will be close to zero. When averaging bipolar values of elementary signals over the row (column) of the MCF containing the amplitude of the signal from a thermal object, mutual compensation will not occur, i.e. arithmetic mean will be significantly different from zero.

Thus, it is possible, by setting the decision threshold close to the arithmetic average calculated by the lines and columns of the MPC, to distinguish between the rows and columns in which the background image is located, from the rows and columns in which the amplitude of the signal from the object is recorded, and also determine their numbers.

The threshold is set each time before starting processing the image array according to the formula (2), using the calculation of the arithmetic mean over the entire image array according to the formula (1). Thus, the changing threshold at each stage of processing for each array adaptively takes into account the average level of energy brightness of a cold extended atmospheric background, since there are several orders of magnitude more elements with background reflections than elements with reflection from a thermal object. With a uniform background in the detection segment, the threshold modulo decreases. In the presence of sections with different brightness, the threshold level p increases. That is, the threshold value adaptively responds to the conditions for detecting thermal objects. A large number of rows and columns of MPCS and a small number of inhomogeneous background regions (darker and less dark) lead to the fact that when averaging over the number of MPCS elements, the threshold p changes insignificantly, however, it responds adaptively to the presence of inhomogeneities in the analyzed segment of the celestial hemisphere. In order for the threshold value p to be less sensitive to the presence of inhomogeneous (different brightness) areas that exist within the detection segment, it is necessary to increase the number of MPCS elements due to the use of more sensitive receivers with more discrete polling.

To test the operability of the proposed method for interperiodic background compensation, mathematical modeling was performed. An exemplary view of the image obtained on the monitor of the video receiving device of the heat finder in the presence of a thermal object, according to which it is proposed to conduct visual detection of thermal objects, is presented in Fig. 3. The image was obtained from the array U _{N, M} , each element u _{n, m of} which contains information about the number of brightness levels (shade of gray) proportional to the brightness of the background and target radiation taken from the cell of the multi-element receiver in the nth line at the mth step . The signal in each cell u _{n, m of} the image array U _{N, M} in this case is quantized into 2 ⁿ levels, where n = 24. It is permissible to use a different number of quantization levels, i.e. the method is operational with n≥8. An array U _{N, M} containing a signal from a point thermal object and an array without a signal from an object used in mathematical modeling are presented in Figs. 4, 5, respectively.

Mathematical modeling included preliminary calculations to calculate the average value μ of the brightness of the background noise of the image array U _{N, M} according to formula (1) and the threshold module calculated by formula (2).

The main part of the simulation included the formation of the MBPC W _{N, M} , in each element with indices n and m, which was recorded the result of multiplication in accordance with formula (3). Variants of arrays W _{N, M} obtained for cases of presence and absence of a thermal object are presented in Figs. 7, 6, respectively.

For arrays W _{N, M} , the average values of μ1 _n were calculated for the columns according to formula (4). Then, μ1 _n values were entered into the corresponding elements of the vector M1. Variants of the Ml vectors obtained for the cases of presence and absence of a thermal object are presented in Figs. 8, 9.

Further, for the arrays W _{N, M} , the average values of μ2 _m were calculated for the columns according to formula (5). Then, the values of μ2 _m were entered into the corresponding elements of the vector M2. Variants of the vectors M2 obtained for the cases of the presence and absence of a thermal object are presented in Figs. 10, 11. The obtained vectors M1, M2 were used to obtain arrays E _{N, M} , in each element having indices n and m, of which the result of multiplication was recorded. in accordance with the formula e _{n, m} = µ1 _n -µ2 _m , where µ1 _n and µ2 _m are the values of the nth and mth elements of the vectors M 1 and M2 after threshold processing, respectively. Variants of arrays E _{N, M} obtained for cases of presence and absence of a thermal object are presented in Figs. 13 and 12, respectively.

в соответствии с правилом:As a result of threshold processing, in which the values of the elements e _{n, m of the} array E _{N, M} were compared with the decision threshold −p _≤ e _{n, m} ≤ p, an output array was obtained

in accordance with the rule:

=0;when the value of the element e _{n, m of the} array E _{N, M} falls into the decision interval [-p, p], the value of the element is reset

= 0;

не попало в интервал принятия решения - р≤e_n,m≤р, ему присваивают единичное значение

=1.if the value of the element is e _{n, m} array

didn’t fall into the decision interval - p≤e _{n, m} ≤р, it is assigned a single value

= 1.

, полученных для случаев наличия и отсутствия теплового объекта, представлены на фиг.15 и 14 соответственно.Output Array Options

obtained for cases of presence and absence of a thermal object are presented in Figs. 15 and 14, respectively.

был использован для создания фоноцелевого изображения на мониторе видеоприемного устройства. Для этого прямоугольный экран монитора видеоконтрольного устройства делили на прямоугольные ячейки по числу строк и столбцов массива E_N,M. Ячейки

, содержащие единичные значения, выделяли цветом, имеющим наибольшую контрастность по отношению к фону экрана монитора. Считали, что в элементах массива со значениями

=1 находились отметки от точечных тепловых объектов (целей). По номерам строк и столбцов элементов, отличающихся по цвету от фона экрана монитора, определяли пространственное положение цели в сегменте полусферы поиска. Изображение на мониторе с обнаруженной точечной тепловой целью представлено на фиг.16.Array

was used to create a phono-target image on the monitor of a video receiving device. For this, the rectangular screen of the monitor of the video monitoring device was divided into rectangular cells by the number of rows and columns of the array E _{N, M.} Cells

containing single values were highlighted with the color having the highest contrast with respect to the background of the monitor screen. It was believed that in the elements of an array with values

= 1 were marks from point thermal objects (targets). By the numbers of rows and columns of elements that differ in color from the background of the monitor screen, the spatial position of the target in the segment of the search hemisphere was determined. The image on the monitor with the detected thermal target is shown in Fig.16.

The method is focused on the detection of thermal objects at long ranges, at which the mark from the air target will occupy one pixel. If you get an even number of point marks from several thermal objects in one row or column, you can skip the target. However, after the completion of the first stage of detection, the target selection process against the sky does not stop, but is repeated many times. The invariable position of a thermal object, in the role of which an aerial target acts, in the same image pixel is an unlikely event. Therefore, skipping a target at one of the stages due to the accidental hit of marks from thermal objects in one row or column does not lead to a complete loss of the target.

Thus, the proposed amplitude method for detecting thermal objects makes it possible to automatically obtain contrast images of thermal objects on the monitor screen without operator intervention, allowing them to determine their angular coordinates, the quantitative composition of the group target, and other characteristics of air targets without first preparing the human operator, which leads to an expansion of the information abilities of a heat finder. In addition, the method allows you to adaptively take into account the differences in energy brightness levels and spatial (angular) sizes of radiation of a small-sized thermal object and a cold extended atmospheric background in any weather situation.

Information sources

1. Yakushenkov Yu.G., Tarasov VV "Looking" type infrared systems. M .: “Logos”, 2004, 430 p.

2. Prikhodko V.N., Khisamov R.Sh. Detection of "point" objects by a direction finder based on a matrix photodetector // Defense Technology, 2007. No. 1-2. S.64-66 (prototype).

3. Allenov A.M., Allenov M.I., Soloviev V.A. Ivanov V.N. Stochastic structure of cloud radiation. St. Petersburg: Gidrometeoizdat, 2000, 175 p.

Claims (1)

A method for detecting thermal objects against the background of the celestial hemisphere, namely, that the optical system of the heat direction finder is guided and then fixed in the segment of the celestial hemisphere of the search, equal to the angular size of the field of view of the optical system of the heat direction finder, in which it is assumed that the thermal object is located, the phonon target image is focused on sensitive areas of the matrix a multi-element radiation detector and, using a lower-case reading circuit, remove an electrical signal proportional to the two-dimensional distribution the energy brightness of the radiation of the background and the thermal object is converted, the signal amplitudes are converted into a digital code using an analog-to-digital converter, the digital code is stored in the random access memory of the video processor in the form of a two-dimensional array U _{N, M} = || u _{n, m} ||, where N is the number of rows, and M is the number of columns of the array, respectively, so that the element of this array u _{n, m} contains information about the voltage proportional to the brightness of the background radiation taken from the cell of the multi-element receiver in the nth row at the mth step, characterized in what number about rows N and the number of columns M of the array U _{N, M} are chosen even, for the image array U _{N, M, the} average value of the background noise brightness is calculated by the formula

determine the module of the threshold value for the decision interval by the formula

, form an array of inter-period background compensation W _{N, M} = || w _{n, m} ||, in each element with indices n and m of which the multiplication result is written in accordance with the formula w _{n, m} = u _{n, m} · (-1) ^{(n + m)} , calculate the average value of the elements of the array W _{N, M} in each row, form a vector M1, the elements of which are the average values calculated for the corresponding rows, calculate the average value of the array W _N , _M in each column, form vector M2, whose elements are average values calculated for the corresponding table ATCs, form two-dimensional array E _{N, M,} in each element with indices n and m which is recorded the result of the multiplication in accordance with formula e _{n, m} = μ1 _n · μ2 _m, where μ1 _n and μ2 _m - value of n-th and m -th elements of the vectors M1 and M2, respectively, for all elements of the array E _{N, M} verify the double inequality -p≤e _{n, m} ≤p and form the output array

as follows: if the element value e _{n, m of the} array E _{N, M} is within the interval [-p, p], then the element value is reset to zero:

= 0, and if the value of the element value e _{n, m of the} array E _{N, M} goes beyond the interval [-p, p], then a unit value is assigned to it:

= 1, when forming a phono-target image on the monitor of a video receiving device, an array is used

why a rectangular monitor screen of a video monitoring device is divided into rectangular cells by the number of rows and columns of the array

cells

containing unit values are highlighted with the color that has the highest contrast with respect to the background of the monitor screen, they decide that in the elements of the array with values

= 1 are marks from point thermal objects, by the numbers of rows and columns of elements that differ in color from the background of the monitor screen, determine the spatial position of thermal objects in the segment of the search hemisphere, in the absence of elements

with single values, they decide on the absence of thermal objects in the analyzed segment of the celestial hemisphere of the search.

Images