CN116049641B - Point target feature extraction method based on infrared spectrum - Google Patents
Point target feature extraction method based on infrared spectrum Download PDFInfo
- Publication number
- CN116049641B CN116049641B CN202310342442.3A CN202310342442A CN116049641B CN 116049641 B CN116049641 B CN 116049641B CN 202310342442 A CN202310342442 A CN 202310342442A CN 116049641 B CN116049641 B CN 116049641B
- Authority
- CN
- China
- Prior art keywords
- target
- data
- spectrum
- infrared
- equivalent
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Abstract
The invention discloses a point target feature extraction method based on infrared spectrum, which comprises the steps of collecting original data, including calibration collection before and after tasks and calibration collection before and after no tasks; and preprocessing the original spectrum data: according to the target time sequence spectrum data and the background data and the four-quadrant fine tracking data, dividing signals and backgrounds in the data, and extracting pure signal data of the target; atmospheric correction processing: the atmospheric parameter measuring system acquires real-time atmospheric parameters, the atmospheric correction software calculates the atmospheric transmittance of the target track, and the target signal after the atmospheric correction is obtained through searching and matching; radiation quantity inversion treatment: the calibration coefficient is acted on the target signal to obtain target spectrum radiation quantity; and (3) quick extraction treatment of infrared characteristics: and (3) equivalent the target to infrared radiation emitted by a plane to obtain a simplified spectral radiation theoretical formula of the target, and obtaining the equivalent temperature and equivalent area characteristics of the target by a rapid processing algorithm.
Description
Technical Field
The invention belongs to the field of target infrared spectrum data processing, and particularly relates to a point target feature extraction method based on infrared spectrum.
Background
For the detection of long-distance indistinguishable point targets, only a few pixels are usually occupied on an image plane; the shape and texture information of the target are lost, the characteristics of the target are difficult to extract only by the gray information and the time sequence information of a few pixels, and the accurate perception of the target is difficult to realize. Far-distance point target infrared multispectral measurement can increase the information content of a detection target in the spectral dimension, and is an important point target non-imaging detection technology. The information such as the target surface temperature, the effective radiation area, the surface material emissivity and the like is mixed in the target infrared spectrum, so that the possibility is brought to the extraction and the identification of the point target characteristics. Wherein readily available equivalent temperatures and equivalent areas, which can represent the overall radiation level of the target, can approximate the temperature and area of the remote point target, whose change over time can reflect the operating state of the target. In general, the equivalent temperature and the equivalent area change slowly, but when the target performs a specific task or fails, the target may have an observable mutation, so that the extraction of the equivalent temperature and the equivalent area features of the point target has important application prospects in enhancing and expanding the perception capability of the remote target.
However, the existing point target feature extraction method based on infrared spectrum has the following problems:
1. the complete flow of feature extraction is not involved: most of algorithms for feature extraction are more involved, and the complete whole flow of target infrared spectrum data acquisition, pretreatment, atmosphere correction, radiation quantity inversion and feature extraction is not elaborated;
2. the extraction precision is poor: the method has the advantages that good noise suppression processing is not carried out on the target data and the background data, so that the accuracy of the inverted target spectrum radiation quantity is poor, and the extracted target feature is poor in accuracy;
3. the stability and the speed of feature extraction can not be considered: the existing feature extraction algorithm can stably extract target features but cannot guarantee the extraction speed, and the Kalman filtering algorithm can rapidly extract but cannot guarantee the stability.
Disclosure of Invention
In order to solve the technical problems, the invention provides a point target feature extraction method based on infrared spectrum, which comprises the steps of acquisition strategy of original data, calibration acquisition before and after tasks and calibration acquisition before and after no tasks; preprocessing of the original spectrum data: the infrared spectrometer collects time sequence spectrum data and background data of a target, and combines four-quadrant fine tracking data to decompose signals and the background in the data, and then the signals are subtracted from the background, so that pure signal data of the target is obtained; atmospheric correction processing: the atmospheric parameter measuring system acquires real-time atmospheric parameters, the atmospheric correction software calculates the atmospheric transmittance of a target track, and the target signal searches the matched atmospheric transmittance and divides the atmospheric transmittance to obtain an atmospheric corrected target signal; radiation quantity inversion treatment: calibrating an infrared radiation spectrometer in advance to obtain a calibration coefficient, and applying the calibration coefficient to a target signal to obtain a target spectrum radiant quantity; and (3) quick extraction treatment of infrared characteristics: the target is equivalent to infrared radiation emitted by a plane, a simplified spectrum radiation theoretical formula of the target can be obtained by a Planck formula, the measured target spectrum data is fitted by the theoretical formula, and the equivalent temperature and equivalent area characteristics of the target are obtained by a rapid processing algorithm.
In order to achieve the above purpose, the invention adopts the following technical scheme:
a point target feature extraction method based on infrared spectrum comprises the following steps:
step A: the infrared spectrum data acquisition comprises two acquisition modes of front and back calibration acquisition before a task and front and back calibration acquisition without a task;
and (B) step (B): preprocessing data;
step C: atmospheric correction processing, namely filtering out the influence of atmospheric transmittance on a target signal; for a certain signal, finding the atmospheric transmittance matched with the pitching and the azimuth of the signal in the transmittance data, thereby obtaining an atmospheric corrected signal;
step D: inversion processing of spectral radiant quantity, obtaining irradiance, radiant intensity and radiant brightness of a target according to definition of radiation calibration, and calculating the irradiance as radiant illuminance E (t):
wherein E (t) is the measured irradiance, alpha is the irradiance response coefficient of the calibration data, S (t) is the target time sequence numerical quantity, B is the matched sky background mean value, and t is the atmospheric transmittance;
step E: the infrared characteristic is rapidly extracted and processed, and the equivalent radiation temperature and the equivalent radiation area of the target are obtained according to the spectral radiation quantity of the target;
step F: and selecting an infrared characteristic extraction strategy, wherein the extraction strategy comprises two strategies of post-hoc and real-time infrared characteristic extraction.
Further, the step B comprises dividing the target spectrum signal and the sky background, counting the mean value B and the mean square error sigma of the sky background, removing coarse errors and obtaining a target pure signal value.
Further, the dividing the target spectrum signal and the sky background comprises processing and distinguishing the target spectrum signal and the sky background by a threshold method; re-confirming by adopting the fine tracking data; for the fine tracking data, the output is 1 when the target is kept up, and the output is 0 when the target is not kept up; when the target data simultaneously satisfies the threshold method and the fine tracking is divided into target signals, the data is finally confirmed as target spectrum signals.
Further, the statistical sky background mean B and the mean square error σ include: for the calibration acquisition before and after the task, the sky background of the same track of the target is acquired before and after the target data acquisition; in order to enable the acquisition of the target pure signal to be more accurate, a sky background section with the same pitch and azimuth of the target signal section is matched and used for calculating a sky background mean value B and a mean square error sigma; for the front-back calibration acquisition without tasks, dividing a plurality of sections of sky background data, calculating the mean value B and the mean square error sigma of each section of sky background, and taking the sky background section with the minimum mean value and the mean square error sum.
Further, the coarse error elimination includes elimination of coarse errors in data, and for the ith frame of data, if the difference between every two adjacent frames of the wave band and the ith frame is smaller than a threshold value Q, the ith frame of data is reserved, otherwise, the ith frame of data is eliminated.
Further, the step E includes: the whole radiation of the target is equivalent to the infrared radiation emitted by a black body with a certain temperature and area, and the equivalent infrared spectrum irradiance of the target at the entrance pupil of the detection equipment is obtained:
wherein ρ is the distance from the target to the detection device, A is the equivalent radiation area, T is the equivalent radiation temperature, and M (T) is the integral of the Planckian function to the response band;
fitting the above formula to actual infrared spectrum data, and constructing an objective function J by using a least square method:
wherein E (T, A) is the calculated target equivalent infrared spectral irradiance,
the target infrared spectrum irradiance is actually measured;
solving a minimized objective function J by using an optimization algorithm to obtain an equivalent radiation temperature T and an equivalent radiation area A;
the optimization algorithm is a modified Gauss-Newton method, and the iteration steps are as follows:
step 1: according to
Beta is a telescoping factor,>
adjusting theoretical spectrum data to the same size scale as actually measured spectrum data; />
Equivalent temperature and equivalent area vector representing the kth calculation,/->
The equivalent infrared spectrum irradiance is the target, k is the iteration number;
step 2: selecting initial data and initial point
Let k=0;
step 3: constructing search directions, and constructing respective search directions by Gauss Newton method
；
Wherein αi is a small amount of diagonal matrix representing the gradient;
step 4: determining a search step size, determining
Is the starting point +.>
Is +.>
Decreasing the objective function value and reversing the method>
；
Step 5: solving a new iteration point to enable
；
Step 6: checking termination condition, judging
If the termination condition is satisfied, stopping the iteration, and outputting an approximately optimal solution>
The method comprises the steps of carrying out a first treatment on the surface of the Otherwise, let k=k+1, go to Step 2.
Further, in the step F, the post-infrared feature extraction is that after the spectrum measurement end and the atmosphere end collect the target spectrum data and the atmosphere data respectively, post-processing is performed on the data, and the equivalent radiation temperature and the equivalent radiation area of the target are extracted; the real-time infrared characteristic extraction is to transmit the target spectrum data and the atmospheric data in real time at the spectrum measuring end and the atmospheric end, and to extract the equivalent radiation temperature and the equivalent radiation area of the target in real time.
Further, in the step a, before and after the task is performed, the sky background data is collected once according to the target track before and after the target data is collected; the front-back calibration without tasks is to collect the target and sky background data by open loop and closed loop segmentation of the fast reflection mirror in the collecting process.
Further, the step D includes:
based on the radiation calibration result, calculating the received spectral radiant quantity according to each parameter of the infrared system;
the spectral irradiance includes irradiance, radiance, and radiant power.
The invention has the advantages that:
(1) Providing a complete whole process of point target infrared feature extraction: the method has the advantages that a full flow of acquisition of target spectrum, sky background and atmospheric data, preprocessing of spectrum data, atmosphere correction, inversion of spectrum radiation quantity and extraction of infrared characteristics is provided, original numerical value data starting from two acquisition strategies of front-back calibration without tasks and front-back calibration with tasks are realized, and equivalent temperature and equivalent area of the target are obtained through processing;
(2) The extraction precision is high: by matching proper sky background data during preprocessing, the precision of a target pure signal is improved, and by searching and matching proper atmospheric transmittance data, the precision of target spectrum data after atmospheric correction is improved, so that the precision of extracting the equivalent temperature and equivalent area of the target is finally improved.
(3) The extraction is stable and the speed is high: according to the characteristics of the target infrared spectrum data, the Gaussian Newton algorithm is improved, the spectrum data is subjected to expansion processing in advance in the algorithm, and a small amount of diagonal matrix is added in the searching direction, so that the stable and rapid extraction of the target infrared characteristics is ensured.
Drawings
FIG. 1 is a flow chart of preprocessing in a pre-and post-task calibration acquisition mode;
FIG. 2 is a flow chart of preprocessing in a non-tasking front-to-back calibration acquisition mode;
FIG. 3 is a timing diagram of the sum of the numerical values of the point target spectrum data;
FIG. 4 is a graph of the results of target data signal and background partitioning;
FIG. 5 is a flow chart of an atmosphere correction process;
FIG. 6 is a spectrum development after atmospheric modification;
FIG. 7 is a graph of the spread spectrum after inversion of the radiation dose and a fitted graph after feature extraction;
FIG. 8 is a modified Gauss-Newton method calculation flow chart;
FIG. 9 is a timing diagram of equivalent temperature and equivalent area after feature extraction;
fig. 10 is a flowchart of a point target feature extraction method based on infrared spectrum according to the present invention.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention. In addition, the technical features of the embodiments of the present invention described below may be combined with each other as long as they do not collide with each other.
As shown in fig. 10, the point target feature extraction method based on infrared spectrum of the present invention includes the following steps:
step A, infrared spectrum data acquisition: the infrared spectrometer collects time sequence spectrum data of a target, and the collection is divided into front and back calibration collection before and after a task and front and back calibration collection before and after no task; the task front-back calibration is to collect once sky background data according to a target track before and after collecting target data; the front-back calibration without tasks is to collect target and background data in a segmented way through an open loop and a closed loop of a quick reflection mirror in the collecting process;
step B, data preprocessing: the data preprocessing mainly comprises the steps of dividing a target data signal and a background, removing coarse errors, denoising and the like, and aims to obtain a smoother signal S and background data, and a mean value B and a mean square error sigma (noise) of a statistical background.
1. Target data signal and background division: because the target light does not enter the infrared radiation spectrometer or the background data is collected manually in an open loop during tracking, the condition that a section of target signal and a section of background appear alternately appears in the collected target data. Therefore, the target data needs to be divided to distinguish the target signal from the sky background. Can be processed by a thresholding method, for the time series target data f (t):
wherein g (t) is a flag for the time-series target data f (t), the data marked 1 is a target valid signal, and the data marked 0 is a background. The key of the method is how to get the threshold H right, where the threshold H is automatically obtained using the following algorithm:
step (1) selecting an average value of the global signal
As an initial estimate;
step (2) averaging the global signals
Dividing the signal, the signal of region G1 being all greater than +.>
The signals of the region G2 are all less than +.>
Is a signal of (2);
step (3) continues to calculate the average value of the areas G1 and G2, respectively
And->
；
Step (4) calculating a new threshold value
；
Step (5) repeating the steps2) Until the stopping condition is satisfied, that is, the threshold difference obtained by two adjacent times of calculation is smaller than the parameter
；
Step (6) outputting the final threshold value
For segmenting data, greater than a final threshold +.>
Is marked as signal, less than the final threshold +.>
Is marked as background.
To further accurately divide the target signal, the fine tracking data may be used for reconfirmation. For fine tracking data, the output is 1 when the target is kept up, and 0 when the target is not kept up. Thus, when the target data simultaneously satisfies the threshold method and the fine tracking is divided into target signals, the data is finally confirmed as the target signals.
2. The mean value B and the mean square error sigma of the sky background are counted: the two acquisition modes of task front-back calibration acquisition and task-free front-back calibration are different in the selection mode of the sky background section. For calibration acquisition before and after a task, sky backgrounds of the same track of a target are acquired before and after target data acquisition. In order to make pure signal acquisition more accurate, sky background sections with the same pitch and azimuth of the target signal section can be matched, and the sky background sections are used for calculating the mean value B and the mean square error sigma of the sky background. For the non-task front-back calibration acquisition, a plurality of sections of sky background data can be usually divided, however, due to incomplete open loop or other reasons, a small amount of target signals are mixed in the background data, so that the number value is higher than that of the background only; therefore, the mean value B and the mean square error sigma of the sky background of each section need to be calculated, and the sky background section with the minimum mean value and the mean square error sum is taken.
3. Rejection of gross errors: coarse errors may occur in the acquired time sequence data, and burrs appear visually on the images, so that extraction of infrared features of the targets is affected. Therefore, coarse errors in the data need to be removed, for the ith frame data, if the difference between two adjacent frames of each band and the ith frame is smaller than a threshold value Q, the ith frame data is reserved, otherwise, the ith frame data is removed, and the Q can generally adopt an empirical value (20 times of sky background mean square value).
4. Obtaining a target pure signal value: the target signal is actually the difference between the target value quantity and the sky background value quantity, so the target pure signal:
s (t) is the divided target time sequence numerical quantity, and B is the matched sky background mean value.
And C, atmosphere correction treatment: the atmospheric correction mainly filters out the influence of atmospheric transmittance on a target signal, and finds out the atmospheric transmittance t matched with the pitch and the azimuth of the signal in transmittance data for a certain signal, thereby obtaining an atmospheric corrected signal (S-B)/t.
Step D, radiation quantity inversion treatment: the dimensional reduction is to invert the radiation quantity signal of the target from the numerical quantity signal, and the irradiance, the radiation intensity, the radiation brightness and the like of the target can be obtained according to the definition of radiation calibration, and are generally calculated as the radiation illuminance E (t):
where E (t) is the measured irradiance and α is the irradiance response coefficient of the calibration data.
And E, rapidly extracting and processing infrared characteristics: and rapidly processing according to the target spectral radiant quantity to obtain the equivalent temperature and equivalent area characteristics of the target.
1. The whole radiation of the target is equivalent to the infrared radiation emitted by a black body with a certain temperature and area, so that the equivalent infrared spectrum irradiance of the target at the entrance pupil of the detection equipment can be obtained:
where ρ is the distance of the target to the detection device, A is the equivalent area, T is the equivalent temperature, and M (T) is the Planckian function integral of the response band.
This formula is used to fit the actual infrared spectrum data, and it is common practice to construct the objective function J by a least squares method:
wherein E (T, A) is the calculated target equivalent infrared spectral irradiance,
is the actual measured target infrared spectral irradiance.
And minimizing the objective function J by using an optimization algorithm to obtain the equivalent radiation temperature T and the equivalent radiation area A.
2. The search direction is the core for the optimization algorithm, and different search directions form different optimization algorithms. Comparing several optimization algorithms, the Gauss-Newton method is found to be the algorithm with the least iteration number and the shortest calculation time. But finds that the method is sensitive to an initial value in the application process, and sometimes leads to iteration solution failure; meanwhile, when the numerical value of the temperature and the area are too large, iterative solution also fails; therefore, the Gauss Newton method is improved, so that the Gauss Newton method is suitable for the problem, and the stability and the wide applicability of the algorithm are improved. As shown in fig. 8, the modified Gauss-Newton method iterates the steps as follows:
Step 1：
beta is a telescoping factor,>
for the theoretical spectral data to be adjusted to the same size scale as the actual measured spectral data;>
equivalent temperature and equivalent area vector representing the kth calculation,/->
For the target equivalent infrared spectral irradiance mentioned above, k is the number of iterations.
Step 2: selecting initial data and initial point
Let k=0.
Step 3: constructing search directions, and constructing respective search directions by Gauss Newton method
。
Wherein, the liquid crystal display device comprises a liquid crystal display device,
for a small number of diagonal matrices +.>
Representing the gradient; which can prevent +.>
The elements in the matrix are excessively reduced, so that the purpose of preventing algorithm degradation is achieved, and alpha is generally 0.05-0.001.
Step 4: determining a search step size, determining
Is the starting point +.>
Is +.>
The objective function value is reduced, and is usually simpleBack-off method makes->
。
Step 5: solving a new iteration point to enable
。
Step 6: checking termination condition, judging
If the termination condition is satisfied, stopping the iteration, and outputting an approximately optimal solution>
The method comprises the steps of carrying out a first treatment on the surface of the Otherwise, let k=k+1, go to Step 2.
Step F, infrared feature extraction strategy: both post-hoc and real-time infrared feature extraction processing strategies may be employed.
1. The post infrared feature extraction is to collect the target spectrum data and the atmosphere data at the spectrum measuring end and the atmosphere end respectively, then to post-process the data, and extract the equivalent radiation temperature and the equivalent radiation area of the target.
2. The real-time infrared characteristic extraction is to transmit the target spectrum data and the atmospheric data in real time at the spectrum measuring end and the atmospheric end, and the equivalent radiation temperature and the equivalent radiation area of the target can be extracted in real time due to the rapid extraction processing capacity of the infrared characteristics.
Embodiments of the present invention are described in detail below with reference to the accompanying drawings.
The invention discloses a point target feature extraction method based on infrared spectrum, which comprises the following specific steps:
step A, data acquisition: calibrating before and after a task, acquiring target spectrum data by a spectrum measuring end, acquiring sky background data of the same track of a target before and after the task, and acquiring atmospheric transmittance data by an atmospheric end; and (3) calibrating before and after no task, wherein a spectrum measuring end opens and closes a loop to acquire target spectrum data and sky background data at intervals in a file, and an atmosphere end acquires atmosphere transmittance data.
Step B, data preprocessing: the method comprises the steps of dividing signals, removing coarse errors and carrying out noise reduction treatment to obtain target pure signal data deducted from the atmospheric background, wherein the preprocessing flows of calibration before and after tasks and calibration before and after no tasks are different, and the flows are shown in the figures 1 and 2.
1. Obtaining final threshold according to the automatic threshold calculation flow
The target data signal is thresholded from the background, and the spectral signal shown in fig. 3 can be thresholded to three signal segments and four background segments as shown in fig. 4.
2. And for calibration acquisition before and after a task, matching sky background sections with the same pitching and azimuth of the target signal section, and calculating the mean value B and the mean square error sigma of the sky background. For the front and back calibration acquisition without tasks, several sections of sky background data are divided, the mean value B and the mean square error sigma of each section of sky background are calculated, and the sky background section with the smallest mean value and the sum of the mean square errors is obtained.
3. And adopting a 20-time sky background mean square value as a coarse error rejection threshold, for certain frame data, if the difference between two adjacent frames of each wave band and the frame is smaller than the threshold, reserving the frame data, otherwise, rejecting the frame data, and carrying out average filtering on the reserved data.
4. And (3) subtracting B from the divided target time sequence numerical value to obtain a matched background mean value according to a formula (6), and obtaining a target pure signal, if the average signal-to-noise ratio of the pure signal is greater than 5, reserving, otherwise, removing the signal.
And C, atmosphere correction treatment: the atmospheric correction mainly filters the influence of atmospheric transmittance on a target signal, and finds the transmittance matched with the pitch and the azimuth of the signal in transmittance data for a certain signal, so that the signal after the atmospheric correction is obtained, the processing flow is shown in fig. 5, and the spectrum expansion after the atmospheric correction is shown in fig. 6.
Step D, radiation quantity inversion treatment: the atmospheric corrected target signal is divided by the irradiance response coefficient to obtain the target measured spectral irradiance, as per equation (7), which is identified in fig. 7.
And E, rapidly extracting and processing infrared characteristics: the equivalent temperature and equivalent area of the target were rapidly extracted using the modified Gauss-Newton method, the calculation flow of which is shown in FIG. 8.
1. The theoretical infrared spectral irradiance of the target at the entrance pupil of the detection device is constructed according to equation (8), and the objective function is constructed by a least squares method according to equation (9).
2. The equivalent temperature and equivalent area of the target were calculated according to the modified Gauss-Newton method calculation procedure, as shown in fig. 9, and the fitted curve to the measured spectrum was identified in fig. 7.
Step F, selecting an infrared feature extraction strategy:
1. the post infrared feature extraction is to collect the target spectrum data and the atmosphere data at the spectrum measuring end and the atmosphere end respectively, then to post-process the data and extract the equivalent temperature and equivalent area of the target.
2. The real-time infrared characteristic extraction is to transmit the target spectrum data and the atmospheric data in real time at the spectrum measuring end and the atmospheric end, and the equivalent temperature and the equivalent area of the target can be extracted in real time due to the rapid extraction processing capacity of the infrared characteristics.
The art-known techniques involved in the present invention are not elaborated in detail. It will be readily appreciated by those skilled in the art that the foregoing description is merely a preferred embodiment of the invention and is not intended to limit the invention, but any modifications, equivalents, improvements or alternatives falling within the spirit and principles of the invention are intended to be included within the scope of the invention.
Claims (4)
1. The point target feature extraction method based on infrared spectrum is characterized by comprising the following steps of:
step one: the infrared spectrum data acquisition comprises two acquisition modes of front and back calibration acquisition before a task and front and back calibration acquisition without a task;
step two: preprocessing data;
step three: atmospheric correction processing, namely filtering out the influence of atmospheric transmittance on a target signal; for a certain signal, finding the atmospheric transmittance matched with the pitching and the azimuth of the signal in the transmittance data, thereby obtaining an atmospheric corrected signal;
step four, inversion processing of the spectrum radiation quantity, obtaining irradiance, radiation intensity and radiation brightness of a target according to definition of radiation calibration, and calculating the irradiance as radiation illuminance E (t):
wherein E (t) is the measured irradiance, alpha is the irradiance response coefficient of the calibration data, S (t) is the target time sequence numerical quantity, B is the matched sky background mean value, and t is the atmospheric transmittance;
step five, rapidly extracting and processing the infrared characteristics, and rapidly processing according to the spectral radiant quantity of the target to obtain the equivalent radiation temperature and equivalent radiation area of the target;
step six, selecting an infrared characteristic extraction strategy, wherein the extraction strategy comprises two strategies of post infrared characteristic extraction and real-time infrared characteristic extraction;
dividing a target spectrum signal and a sky background, counting a mean value B and a mean square error sigma of the sky background, removing coarse errors and obtaining a target pure signal value;
dividing the target spectrum signal and the sky background comprises adopting a threshold method to process and distinguish the target spectrum signal and the sky background; re-confirming by adopting the fine tracking data; for the fine tracking data, the output is 1 when the target is kept up, and the output is 0 when the target is not kept up; when the target data simultaneously meets a threshold method and the fine tracking is divided into target signals, finally confirming the data as target spectrum signals;
the statistical sky background mean value B and the mean square error sigma comprise: for the calibration acquisition before and after the task, the sky background of the same track of the target is acquired before and after the target data acquisition; in order to enable the acquisition of the target pure signal to be more accurate, a sky background section with the same pitch and azimuth of the target signal section is matched and used for calculating a sky background mean value B and a mean square error sigma; for the front-back calibration acquisition without tasks, dividing a plurality of sections of sky background data, calculating a sky background mean value B and a mean square error sigma of each section, and taking a sky background section with the smallest mean value and the sum of the mean square errors;
the coarse error elimination comprises the steps of eliminating coarse errors in data, and for the ith frame of data, if the difference between two adjacent frames of each wave band and the ith frame is smaller than a threshold value Q, reserving the ith frame of data, otherwise eliminating the ith frame of data;
the fifth step comprises the following steps: the whole radiation of the target is equivalent to the infrared radiation emitted by a black body with a certain temperature and area, so that the equivalent theoretical infrared spectrum irradiance at the entrance pupil of the detection equipment is obtained:
wherein ρ is the distance from the target to the detection device, A is the equivalent radiation area, T is the equivalent radiation temperature, and M (T) is the integral of the Planckian function to the response band;
fitting the above formula to actual infrared spectrum data, and constructing an objective function J by using a least square method:
wherein J is an objective function, E (T, A) is a calculated target equivalent infrared spectral irradiance,
the target infrared spectrum irradiance is actually measured;
solving a minimized objective function J by using an optimization algorithm to obtain an equivalent radiation temperature T and an equivalent radiation area A;
the optimization algorithm is a modified Gauss-Newton method, and the iteration steps are as follows:
step 1: according to
Beta is a telescoping factor,>
adjusting theoretical spectrum data to the same size scale as actually measured spectrum data; />
Equivalent temperature and equivalent area vector representing the kth calculation,/->
The infrared spectrum irradiance is the target theory, and k is the iteration number;
step 2: selecting initial data and initial point
Let k=0;
step 3: constructing search directions, and constructing respective search directions by Gauss Newton method
；
Wherein, the liquid crystal display device comprises a liquid crystal display device,
for a small number of diagonal matrices +.>
Representing the gradient;
step 4: determining a search step size, determining
Is the starting point +.>
Is +.>
The objective function value is reduced, and after useBack-treatment of->
；
Step 5: solving a new iteration point to enable
；
Step 6: checking termination condition, judging
If the termination condition is satisfied, stopping the iteration, and outputting an approximately optimal solution>
The method comprises the steps of carrying out a first treatment on the surface of the Otherwise, let k=k+1, go to Step 2.
2. The method for extracting point target features based on infrared spectrum according to claim 1, wherein in the sixth step, the post-infrared feature extraction is to perform post-processing on the data after the spectrum measurement end and the atmosphere end collect the target spectrum data and the atmosphere data respectively, so as to extract the equivalent radiation temperature and the equivalent radiation area of the target; the real-time infrared characteristic extraction is to transmit the target spectrum data and the atmospheric data in real time at the spectrum measuring end and the atmospheric end, and to extract the equivalent radiation temperature and the equivalent radiation area of the target in real time.
3. The method for extracting point target features based on infrared spectrum according to claim 1, wherein in the first step, the calibration before and after the task is to collect sky background data once according to the target track before and after the target data is collected; the front-back calibration without tasks is to collect the target and sky background data by open loop and closed loop segmentation of the fast reflection mirror in the collecting process.
4. The method for extracting point target features based on infrared spectrum according to claim 1, wherein the fourth step comprises:
based on the radiation calibration result, calculating the received spectral radiant quantity according to each parameter of the infrared system;
the spectral irradiance includes irradiance, radiance, and radiant power.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310342442.3A CN116049641B (en) | 2023-04-03 | 2023-04-03 | Point target feature extraction method based on infrared spectrum |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310342442.3A CN116049641B (en) | 2023-04-03 | 2023-04-03 | Point target feature extraction method based on infrared spectrum |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116049641A CN116049641A (en) | 2023-05-02 |
| CN116049641B true CN116049641B (en) | 2023-06-30 |
Family
ID=86133565
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310342442.3A Active CN116049641B (en) | 2023-04-03 | 2023-04-03 | Point target feature extraction method based on infrared spectrum |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116049641B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105784626A (en) * | 2016-04-05 | 2016-07-20 | 中国科学院合肥物质科学研究院 | Atmospheric pollutant self-adaptive identification method and system based on infrared spectrum imaging technology |
| CN112484846A (en) * | 2020-11-12 | 2021-03-12 | 北京环境特性研究所 | Method and device for measuring radiation illumination of air point target by using infrared band |
Family Cites Families (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP6585049B2 (en) * | 2013-08-28 | 2019-10-02 | ランダー オーディオ インコーポレイテッド | System and method for automatic audio generation using semantic data |
| CN103776540B (en) * | 2013-12-30 | 2015-07-08 | 华中科技大学 | Multiband common-optical-path spectrum combined remote sensing measurement system and method thereof |
| CN105203465B (en) * | 2015-09-16 | 2018-03-16 | 湖北久之洋红外系统股份有限公司 | A kind of ultraphotic spectrum infrared-imaging gas monitoring device and its monitoring method |
| CN105607251B (en) * | 2015-12-31 | 2017-03-22 | 南京华图信息技术有限公司 | Multi-mode infrared imaging system and method facing weak target detection |
| CN106018332A (en) * | 2016-07-21 | 2016-10-12 | 华南农业大学 | Near-infrared-spectrum citrus yellow shoot disease field detection method |
| CN107918937B (en) * | 2017-12-06 | 2021-07-30 | 电子科技大学 | Target and background physical superposition method based on spectral radiation |
| CN110837130B (en) * | 2019-11-22 | 2021-08-17 | 中国电子科技集团公司第四十一研究所 | Target automatic detection algorithm based on millimeter wave/terahertz wave radiation |
| CN114707595B (en) * | 2022-03-29 | 2024-01-16 | 中国科学院精密测量科学与技术创新研究院 | Spark-based hyperspectral laser radar multichannel weighting system and method |
| CN115265803A (en) * | 2022-07-27 | 2022-11-01 | 中国科学院西安光学精密机械研究所 | Method for measuring infrared spectrum radiation characteristics of remote point target |
| CN115615560A (en) * | 2022-10-26 | 2023-01-17 | 中国科学院光电技术研究所 | Wide dynamic range high precision infrared radiation measuring method |
-
2023
- 2023-04-03 CN CN202310342442.3A patent/CN116049641B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105784626A (en) * | 2016-04-05 | 2016-07-20 | 中国科学院合肥物质科学研究院 | Atmospheric pollutant self-adaptive identification method and system based on infrared spectrum imaging technology |
| CN112484846A (en) * | 2020-11-12 | 2021-03-12 | 北京环境特性研究所 | Method and device for measuring radiation illumination of air point target by using infrared band |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116049641A (en) | 2023-05-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Finoguenov et al. | The XMM-Newton wide-field survey in the COSMOS field: statistical properties of clusters of galaxies | |
| CN106940887B (en) | GF-4 satellite sequence image cloud and cloud shadow detection method | |
| CN107944368B (en) | Water body extraction method based on time sequence remote sensing image | |
| EP2360642A2 (en) | Video object tracking | |
| CN109559324A (en) | A kind of objective contour detection method in linear array images | |
| CN111007039B (en) | Automatic extraction method and system for sub-pixel level water body of medium-low resolution remote sensing image | |
| CN106296613B (en) | A kind of Dual Energy Subtraction method based on DR machine | |
| CN109359604B (en) | Method for identifying instrument under shadow interference facing inspection robot | |
| CN108510544B (en) | Light strip positioning method based on feature clustering | |
| Moralejo et al. | MARS, the MAGIC analysis and reconstruction software | |
| CN108921170A (en) | A kind of effective picture noise detection and denoising method and system | |
| CN111145198A (en) | Non-cooperative target motion estimation method based on rapid corner detection | |
| CN116049641B (en) | Point target feature extraction method based on infrared spectrum | |
| CN106204596B (en) | Panchromatic waveband remote sensing image cloud detection method based on Gaussian fitting function and fuzzy mixed estimation | |
| CN109784229B (en) | Composite identification method for ground building data fusion | |
| CN110929739A (en) | Automatic impervious surface range remote sensing iterative extraction method | |
| CN109946670A (en) | A kind of polarization radar information extracting method of optical image driving | |
| Fischer et al. | Median spectral-spatial bad pixel identification and replacement for hyperspectral SWIR sensors | |
| CN113552648A (en) | Astronomical positioning and automatic tracking method and system for space target and electronic equipment | |
| CN111089607B (en) | Automatic calibration method for detection capability of telescope system | |
| CN109726641B (en) | Remote sensing image cyclic classification method based on automatic optimization of training samples | |
| US20230245445A1 (en) | An object detection method | |
| CN110349173B (en) | Ground feature change monitoring method based on high-resolution remote sensing image | |
| CN111062954A (en) | Infrared image segmentation method, device and equipment based on difference information statistics | |
| CN116468671B (en) | Plant disease degree detection method, device, electronic apparatus, and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |