rus
Issue #3/2023
F. G. Agayev, H. H. Asadov, G. V. Alieva
Narrow-band Signature Dual-wavelength Method for Photometric Control and Detection of Well-known Configuration Objects Against the Vegetation Background
Narrow-band Signature Dual-wavelength Method for Photometric Control and Detection of Well-known Configuration Objects Against the Vegetation Background
DOI: 10.22184/1993-7296.FRos.2023.17.3.232.237
The theoretical fundamentals of the proposed dual-wavelength narrow-band method of photometric control and detection of the well-known configuration target against the forest background are outlined. The target is detected by calculating its NIR- and RED- narrowband signatures using the proposed formulas at the wavelength intervals with a width of Δλ, closely adjacent to the “red edge” area in the vegetation reflection spectrum and further comparing these signatures with the relevant vegetation signatures. The proposed method provides for the availability of a priori data on the forest spectral signatures, as well as on the ratio of the forest areas and the target on the generated image.
The theoretical fundamentals of the proposed dual-wavelength narrow-band method of photometric control and detection of the well-known configuration target against the forest background are outlined. The target is detected by calculating its NIR- and RED- narrowband signatures using the proposed formulas at the wavelength intervals with a width of Δλ, closely adjacent to the “red edge” area in the vegetation reflection spectrum and further comparing these signatures with the relevant vegetation signatures. The proposed method provides for the availability of a priori data on the forest spectral signatures, as well as on the ratio of the forest areas and the target on the generated image.
Теги: detection forests narrow-band method photometric control red edge signature красный переход леса обнаружение сигнатура узкоспектральный метод фотометрический контроль
Narrow-band Signature
Dual-wavelength Method for Photometric Control and Detection of
Well-known Configuration Objects Against the Vegetation Background
F. G. Agayev, H. H. Asadov, G. V. Alieva
National Aerospace Agency, Baku, Republic of Azerbaijan
The theoretical fundamentals of the proposed dual-wavelength narrow-band method of photometric control and detection of the well-known configuration target against the forest background are outlined. The target is detected by calculating its NIR- and RED- narrowband signatures using the proposed formulas at the wavelength intervals with a width of Δλ, closely adjacent to the “red edge” area in the vegetation reflection spectrum and further comparing these signatures with the relevant vegetation signatures. The proposed method provides for the availability of a priori data on the forest spectral signatures, as well as on the ratio of the forest areas and the target on the generated image.
Keywords: signature, red edge, forests, detection, narrow-band method, photometric control
Received: 30.01.2023
Accepted: 24.02.2023
Introduction
The problem of photometric control and detection of objects against the forest background using the hyperspectral technology was solved in a number of papers, for example, in [1, 2]. The solution of this problem is relevant in order to ensure the security issues, search and rescue problems, arrangement of protection against the border violators, detection of the enemy objects during the military conflicts, etc. Moreover, as noted in [2], the solution of this problem is rather difficult, mainly due to the high degree of background variability, including the vegetation composition, background textures, and lighting conditions. A special case of this problem is the detection of objects with a well-known surface area against the forest background. Such objects can be people, various animals, cars, military equipment, etc., the surface area of which is known a priori. According to [3], the background objects in the general case not only have an emission factor and thermodynamic temperature, but also contain such effects as the sky and sun radiation reflection, influence of the earth’s surface temperature.
The general circuit for remote determination of such ground object indicators is as follows. In a wide range of wavelengths, the following problem of minimizing the difference ∆ should be solved using the least squares method:
, (1)
where λ is the wavelength;
εg is the emission factor of the “gray” target;
Tg is the temperature of the same target;
Lmeans.cor is an atmospheric-adjusted measured radiation. The last value is determined as follows:
, (2)
where Lmeans(λ) is the measured radiation from the object;
Latm(λ) is the initial atmospheric radiation;
τ(λ) is the atmospheric optical thickness.
Thus, the procedure (1)–(2) makes it possible to assimilate the photometrically controlled object to the gray body model by selecting εg, Tg.
Moreover, such a general approach to the identification of various objects as the gray bodies obviously does not allow to solve the object identification problem that are against a background with a complex reflection spectral-response characteristic, such as an agglomeration of trees or a forest area.
Vegetation, and in particular the trees, has a specific reflection spectrum with several interesting narrow-band regions that allow to use the narrow-band signatures to detect objects located against the forest background.
A number of papers are devoted to the theory and practice of using narrow-band signatures (for example, [4–6]). According to [4], the application of narrow-band signatures is certainly useful if the source emits or reflects radiation in a narrow wavelength range. However, the narrow-band spectral signatures can also be useful if some a priori relations between these signatures are well-known. As for the original problem of object detection and identification against the forest background, let us consider the possible formation of narrow-band signatures.
As noted in [7], the main feature of the vegetation reflection spectrum is the red edge effect given in Fig. 1.
According to [7], chlorophyll available in the plants strongly absorbs electromagnetic radiation in the blue (400–500 nm) and red (600–700 nm) regions. In the near infrared region, the pigments absorb radiation on a negligible level, resulting in the formation of a red edge region in the wavelength range of 700–730 nm. Moreover, the red edge effect has a multipurpose nature for vegetation. However, the “red edge” steepness depends on a number of factors, such as the plant type, phenology, and availability of the plant stress. Further, this article describes the proposed method for photometric control and detection of various objects, with a well-known surface area, located against the forest background, with the use of narrow-band spectral signatures, the spectral regions of which are closely adjacent to the region λ1 ÷ λ2 on the right and on the left with a width Δλ (Fig. 1).
Photometric control method
Let us assume that we have an optical system that generates an object image (Fig. 2). Its spectral range covers the wavelength range [(λ1 − Δλ) ÷ (λ1 + Δλ)]. We believe that the measurements can be made in the RED and NIR regions with the width ∆λ shown in Fig. 1 using the detector. Moreover, the type and properties of the background vegetation are as follows, in particular, the η ratio, determined as follows:
, (3)
where LABCD(NIR) is radiation in the NIR range, coming from the forest, in the absence of objects; LABCD(RED) is radiation in the RED range, coming from the forest, in the absence of an object.
We should note that the ratio η is either a priori known or determined experimentally, when measured in the NIR and RED ranges in that part of the forest where there are no objects against the vegetation background. We know the nature of the object to be detected, and we also know the surface area of the search target. Then we can propose the following object detection algorithm:
After measuring the total flow SNIR in the NIR range, we will determine the following:
, (4)
where ; is a well-known value; the “0b” ratio indicates belonging to the object.
By measuring the total flow SRED in the RED range, we will find the following:
. (5)
We will get the following using (3) and (4):
. (6)
We will get the following using (5) and (6):
. (7)
With due regard to (4) and (5), we will obtain the following:
(8)
Thus, (7) and (8) represent the combined equations, where LΣ (forest) is an additionally measured forest parameter in the NIR and RED ranges, in that part of the forest where there is a confirmed absence of any background objects.
The solution of the combined equations (7), (8) is obtained in the following form:
, (9)
, (10)where:
,
.
Thus, the proposed method of dual-wavelength narrow-band signatures makes it possible to control and detect the targets with a well-known surface area, using their photometric images against the forest background by calculating the narrow-band signature values of the search objects and comparing these values with the relevant forest indicators. The detection criterion can be fulfillment of the following conditions:
, (11)
. (12)
Conclusion
The dual-wavelength narrow-band method for photometric control and detection of targets with a well-known surface area against the forest background is proposed. The target detection criterion is the mismatch between the values of the NIR and RED narrow-band signatures, the forest, and the target. The values are calculated at the wavelength intervals with a width of Δλ, closely adjacent to the “red edge” region in the vegetation reflection spectrum according to the developed formulas. The proposed method provides for the availability of a priori data on the forest spectral signatures, as well as on the ratio of the forest and target areas on the generated photometric image. The analytical formulas have been obtained for calculating the narrow-band spectral signatures of the search targets used for their control, detection and identification against the forest background.
AUTHORS
Agayev Fakhraddin Gyulali ogly, Dr. of technical sciences, professor, National Aerospace Agency, Baku, Republic of Azerbaijan.
ORCID 0000-0002-9826-0868
Asadov Hikmet Hamid ogly, Dr. of technical sciences, professor, National Aerospace Agency, Baku, Republic of Azerbaijan.
ORCID 0000-0003-1180-1535
Alieva Gunel Vagif gyzy, Cand. of technical sciences, National Aerospace Agency, Baku, Republic of Azerbaijan.
ORCID 0000-0001-6540-8750
Dual-wavelength Method for Photometric Control and Detection of
Well-known Configuration Objects Against the Vegetation Background
F. G. Agayev, H. H. Asadov, G. V. Alieva
National Aerospace Agency, Baku, Republic of Azerbaijan
The theoretical fundamentals of the proposed dual-wavelength narrow-band method of photometric control and detection of the well-known configuration target against the forest background are outlined. The target is detected by calculating its NIR- and RED- narrowband signatures using the proposed formulas at the wavelength intervals with a width of Δλ, closely adjacent to the “red edge” area in the vegetation reflection spectrum and further comparing these signatures with the relevant vegetation signatures. The proposed method provides for the availability of a priori data on the forest spectral signatures, as well as on the ratio of the forest areas and the target on the generated image.
Keywords: signature, red edge, forests, detection, narrow-band method, photometric control
Received: 30.01.2023
Accepted: 24.02.2023
Introduction
The problem of photometric control and detection of objects against the forest background using the hyperspectral technology was solved in a number of papers, for example, in [1, 2]. The solution of this problem is relevant in order to ensure the security issues, search and rescue problems, arrangement of protection against the border violators, detection of the enemy objects during the military conflicts, etc. Moreover, as noted in [2], the solution of this problem is rather difficult, mainly due to the high degree of background variability, including the vegetation composition, background textures, and lighting conditions. A special case of this problem is the detection of objects with a well-known surface area against the forest background. Such objects can be people, various animals, cars, military equipment, etc., the surface area of which is known a priori. According to [3], the background objects in the general case not only have an emission factor and thermodynamic temperature, but also contain such effects as the sky and sun radiation reflection, influence of the earth’s surface temperature.
The general circuit for remote determination of such ground object indicators is as follows. In a wide range of wavelengths, the following problem of minimizing the difference ∆ should be solved using the least squares method:
, (1)
where λ is the wavelength;
εg is the emission factor of the “gray” target;
Tg is the temperature of the same target;
Lmeans.cor is an atmospheric-adjusted measured radiation. The last value is determined as follows:
, (2)
where Lmeans(λ) is the measured radiation from the object;
Latm(λ) is the initial atmospheric radiation;
τ(λ) is the atmospheric optical thickness.
Thus, the procedure (1)–(2) makes it possible to assimilate the photometrically controlled object to the gray body model by selecting εg, Tg.
Moreover, such a general approach to the identification of various objects as the gray bodies obviously does not allow to solve the object identification problem that are against a background with a complex reflection spectral-response characteristic, such as an agglomeration of trees or a forest area.
Vegetation, and in particular the trees, has a specific reflection spectrum with several interesting narrow-band regions that allow to use the narrow-band signatures to detect objects located against the forest background.
A number of papers are devoted to the theory and practice of using narrow-band signatures (for example, [4–6]). According to [4], the application of narrow-band signatures is certainly useful if the source emits or reflects radiation in a narrow wavelength range. However, the narrow-band spectral signatures can also be useful if some a priori relations between these signatures are well-known. As for the original problem of object detection and identification against the forest background, let us consider the possible formation of narrow-band signatures.
As noted in [7], the main feature of the vegetation reflection spectrum is the red edge effect given in Fig. 1.
According to [7], chlorophyll available in the plants strongly absorbs electromagnetic radiation in the blue (400–500 nm) and red (600–700 nm) regions. In the near infrared region, the pigments absorb radiation on a negligible level, resulting in the formation of a red edge region in the wavelength range of 700–730 nm. Moreover, the red edge effect has a multipurpose nature for vegetation. However, the “red edge” steepness depends on a number of factors, such as the plant type, phenology, and availability of the plant stress. Further, this article describes the proposed method for photometric control and detection of various objects, with a well-known surface area, located against the forest background, with the use of narrow-band spectral signatures, the spectral regions of which are closely adjacent to the region λ1 ÷ λ2 on the right and on the left with a width Δλ (Fig. 1).
Photometric control method
Let us assume that we have an optical system that generates an object image (Fig. 2). Its spectral range covers the wavelength range [(λ1 − Δλ) ÷ (λ1 + Δλ)]. We believe that the measurements can be made in the RED and NIR regions with the width ∆λ shown in Fig. 1 using the detector. Moreover, the type and properties of the background vegetation are as follows, in particular, the η ratio, determined as follows:
, (3)
where LABCD(NIR) is radiation in the NIR range, coming from the forest, in the absence of objects; LABCD(RED) is radiation in the RED range, coming from the forest, in the absence of an object.
We should note that the ratio η is either a priori known or determined experimentally, when measured in the NIR and RED ranges in that part of the forest where there are no objects against the vegetation background. We know the nature of the object to be detected, and we also know the surface area of the search target. Then we can propose the following object detection algorithm:
After measuring the total flow SNIR in the NIR range, we will determine the following:
, (4)
where ; is a well-known value; the “0b” ratio indicates belonging to the object.
By measuring the total flow SRED in the RED range, we will find the following:
. (5)
We will get the following using (3) and (4):
. (6)
We will get the following using (5) and (6):
. (7)
With due regard to (4) and (5), we will obtain the following:
(8)
Thus, (7) and (8) represent the combined equations, where LΣ (forest) is an additionally measured forest parameter in the NIR and RED ranges, in that part of the forest where there is a confirmed absence of any background objects.
The solution of the combined equations (7), (8) is obtained in the following form:
, (9)
, (10)where:
,
.
Thus, the proposed method of dual-wavelength narrow-band signatures makes it possible to control and detect the targets with a well-known surface area, using their photometric images against the forest background by calculating the narrow-band signature values of the search objects and comparing these values with the relevant forest indicators. The detection criterion can be fulfillment of the following conditions:
, (11)
. (12)
Conclusion
The dual-wavelength narrow-band method for photometric control and detection of targets with a well-known surface area against the forest background is proposed. The target detection criterion is the mismatch between the values of the NIR and RED narrow-band signatures, the forest, and the target. The values are calculated at the wavelength intervals with a width of Δλ, closely adjacent to the “red edge” region in the vegetation reflection spectrum according to the developed formulas. The proposed method provides for the availability of a priori data on the forest spectral signatures, as well as on the ratio of the forest and target areas on the generated photometric image. The analytical formulas have been obtained for calculating the narrow-band spectral signatures of the search targets used for their control, detection and identification against the forest background.
AUTHORS
Agayev Fakhraddin Gyulali ogly, Dr. of technical sciences, professor, National Aerospace Agency, Baku, Republic of Azerbaijan.
ORCID 0000-0002-9826-0868
Asadov Hikmet Hamid ogly, Dr. of technical sciences, professor, National Aerospace Agency, Baku, Republic of Azerbaijan.
ORCID 0000-0003-1180-1535
Alieva Gunel Vagif gyzy, Cand. of technical sciences, National Aerospace Agency, Baku, Republic of Azerbaijan.
ORCID 0000-0001-6540-8750
Readers feedback
